- Email: [email protected]
Measurement of Anticomplementary Activity in Therapeutic Intravenous Immunoglobulin Preparations
827674—BIOL 25/1 (ISSUE)—MS 0017
Biologicals (1997) 25, 87–92
Measurement of Anticomplementary Activity in Therapeutic Intravenous Immunoglobulin Preparations Indra Ramasamy,* Em Tran and Albert Farrugia Molecular Biology, Therapeutic Goods Administration Laboratory, P.O. Box 100, Woden, Canberra, ACT 2606, Australia
Abstract. Anticomplementary activity (ACA) of aggregates in intravenous immunoglobulin preparations (IVIG) was investigated using the modified Kabat and Meyer classical complement consumption method recommended by the European Pharmacopoeia and a C1q-coated microtitre enzyme-linked immunosorbent assay (ELISA). The physical characteristics of aggregates were found to affect complement binding. Aggregates formed by heating IVIG preparations at acid pH bound complement poorly, while aggregates formed by heating IVIG at neutral pH showed high ACA. This suggests that analysis of complement binding capacity provides a level of aggregate characterization of aggregates which is additional to quantitation by High-performance liquid chromatography (HPLC). The correlation (r = 0·98) between the two tests was good when aggregates formed at neutral pH were compared, but decreased (r = 0·57) when aggregates formed at acid pH were included. A comparison of the results showed that there were no significant differences in the classification of aggregates with acceptable/unacceptable (i.e. pass/fail outcome) values of ACA. Between assay variation (CV = 7·6%) was lower in the ELISA test compared with the complement consumption assay (where percentage binding varied from 78·9% to 100%). Both assays are justified for the evaluation of ACA in therapeutic IVIG. The ELISA had the advantage in being more precise, less dependent on reagent source and requiring less technical expertise. = 1997 The International Association of Biological Standardization
Introduction Early attempts to administer immunoglobulin (Ig) preparations intravenously to patients with primary immunodeficiency were associated with severe anaphylaxis type reactions.1 The exact causes of such adverse reactions are not known, although a variety of theories have been put forward to explain them. The presence of aggregated IgG have been proposed as a cause of such reactions, particularly if these aggregates possess complement binding activity. The measurement of anticomplementary activity in immunoglobulin preparations is therefore an important aspect of their pharmaceutical quality control. A method based on complement mediated lysis of antibody-coated sheep red blood cells (SRBC) was adopted as the reference method for measuring *To whom correspondence should be addressed: Dr. I. Ramasamy, Molecular Biology, Therapeutic Goods Administration Laboratory, P.O. Box 100, Woden, ACT 2606, Canberra, Australia. 1045–1056/97/010087 + 6 $25.00/0/bg960063
ACA by the European Pharmacopeia (EP) in 1995.2,3 The method is cumbersome and not amenable to screening large numbers of samples. However, the need to measure ACA in Ig preparations has become more acute in recent years, due to the introduction of viral inactivation steps in the manufacture of plasma products. These include heat treatment at 60°C, which may generate aggregate formation and ACA. The tendency to form aggregates is further influenced by product formulation, and storage time and conditions. As manufacturers introduce viral inactivation steps under pressure from regulatory authorities and product consumers,4,5 the monitoring of the various manufacturing processes and each individual product for ACA becomes relevant. An alternative technique for measuring ACA uses an enzyme-linked immunosorbent assay principle. Aggregates in the Ig preparation are captured onto microtitre plate wells coated with C1q, the first step in the classical pathway for complement mediated lysis. Detection of bound aggregate is 7 1997 The International Association of Biological Standardization
BIO 25/1 ISSUE (827674)L—MS 0017
88
Ramasamy et al .
by horseradish peroxidase-conjugated protein A. The method’s initial optimization, using in house reagents, was for the measurement of circulating immunecomplexes in serum,6,7 but publications have shown the feasibility of using it to measure ACA in IVIG preparations.8 A problem with this approach is that investigators have used arbitrary units of measurement. However, the principle of biological standardization requires the availability of an appropriately calibrated standard, and the measurement of ACA in international units assigned through the reference method. This report is the first to compare the use of the EP and ELISA methods in measuring ACA in immunoglobulin preparations. Both methods were optimised for routine use, and aggregates formed under different conditions were then analysed.
Materials and methods Preparations The European Pharmacopoeia Ig preparation (50 g/l, pH 7⋅0), received as a lyophilized powder, from European Pharmacopoiea Commission (EPC) was used as both a negative control and at a higher concentration a positive control, as recommended by the EPC. Four batches of commercial IgG for intravenous use (Intragam, formulated at 60 g/l at pH 4⋅2) were obtained from CSL Bioplasma (Vic, Australia). The product is manufactured using Cohn fractionation followed by low pH incubation to suppress aggregate formation.9 Preparation of aggregated immunoglobulin Intragam was adjusted to pH 7⋅0 with 5 n NaOH, a concentrated solution of NaOH was used to minimize the change in volume. The reconstituted reference preparation, and Intragam at both acid (pH 4⋅2) and neutral pH (pH 7⋅0), were heated at 61°C in a water bath for various times and cooled to 4°C. The percentage aggregate formed was measured by HPLC gel filtration chromatography on a 0⋅6 m × 7⋅5 mm TSK gel G 3000SW column (ToSoh, Japan) as described by the European Pharmacopoiea.10 In the chromatogram the principal peak, molecular weight approximately 160 000, corresponded to the IgG monomer, and there were two further peaks corresponding to the dimer and aggregates. Low molecular weight degradation products were not observed. An increased rate of aggregate formation was found at acid pH compared with neutral pH.
Assay protocols 1. Complement binding analysis: Complement mediated lysis of antibody coated sheep red blood cells. The assay procedure was carried out according to the European Pharmacopoiea.2 Immunoglobulin preparation (10 mg) under investigation was incubated with guinea-pig complement in gelatin– barbitone buffer in 1 ml of reaction mixture. The titre of the remaining complement was measured by the lysis of rabbit antibody coated sheep red blood cells (Bio-Whittaker, USA). Anticomplementary activity of the immunoglobulin preparation is calculated as the percentage consumption of complement relative to the total complement in the reaction mix. Unless otherwise stated fresh guinea-pig complement was used; for comparison purposes lyophilized guinea-pig complement was obtained from two sources (BioMerieux, France and BioWhittaker). The haemolytic unit of complement activity (CH50 ) is the amount of complement that, under given reaction conditions will produce the lysis of 2⋅5 × 108 cells out of a total of 5 × 108 optimimally sensitized red blood cells. 2. Complement binding analysis: enzyme-linked immunosorbent assay. Flat bottomed microtitre plates (Immunlon, Dynatech) were coated overnight at 4°C with 100 ml of human C1q (Sigma Chemicals, St Louis, MO, USA) in 0⋅05 m bicarbonate/carbonate buffer pH 9⋅6. Plates were washed twice for 10 min in PBS–Tween and followed by a further wash for 1 h, the latter served as a blocking step. PBS–Tween was made up in 1 litre of distilled water containing 8⋅0 g NaCl, 0⋅2 g potassium dihydrogen phosphate, 1⋅15 g disodium hydrogen phosphate, 0⋅2 g potassium chloride, 1 g bovine serum albumin, 1 ml Tween-20, pH 7⋅4. Varying concentrations of test sample in 100 ml PBS were added to the microtitre plate and incubated for 3 h at 37°C. Following 3 further washes for 15 min, the plates were incubated for 1⋅5 h at 37°C with 100 ml of horseradish peroxidase-conjugated protein A (Sigma Chemicals) diluted in PBS–Tween. The plates were washed three times and 100 ml of substrate solution made up freshly with 40 mg o-phenylenediamine (OPD) and 40 ml of 30% hydrogen peroxide in 24⋅3 mls 0⋅1 m citric acid, 25⋅7 ml 0⋅2 m disodium hydrogen phosphate and 50 ml water, pH 5⋅0, was added. After incubation in the dark for 15 min at 37°C the reaction was stopped with 50 ml of 2⋅5 m sulphuric acid. The light absorption was read at 492 nm in a microplate reader (Titertrek,
BIO 25/1 ISSUE (827674)L—MS 0017
Anticomplementary activity of intravenous immunoglobulins
Flow laboratories, U.K.). When a standard curve was constructed curve fitting and analyte concentration interpolation were performed using a polynomial third order function.
Table 2. EP reference method: binding of aggregates to complement—variation with complement levels in incubation mix
RESULTS
Complement concentration (CH50 units)
Complement binding (CH 50/mg)
4⋅50 9⋅00 17⋅54
0⋅228 0⋅378 0⋅564
European Pharmacopoeia reference method Factors that affect complement binding by aggregates. Table 1 shows the variation of anticomplementary activity with amount of aggregate in the reaction mixture. Under defined conditions each individual aggregate preparation showed a characteristic maximum binding; an increased maximum binding of complement was found in preparations with higher concentrations of aggregate. This suggests that, as well as the amount of aggregate generated, the physical properties of the aggregates are influenced by the length of time the IVIG was heated, with a consequent effect on the maximum potential binding of each aggregate. In further experiments anticomplementary activity increased proportionately with complement levels in the incubation mix (Table 2). Standard curve. Figure 1 shows the relationship between aggregate concentration and complement binding of different preparations. At aggregate levels greater than 5%, complement binding plateaued; however, a linear relationship was observed at lower aggregate levels with aggregates formed at neutral pH. Aggregates formed at acid pH were incapable of binding complement at all aggregate levels tested, suggesting denaturation of the Fc region.
89
Reference immunoglobulin preparation (10 mg) with 4⋅35% aggregate concentration was incubated with varying amounts of complement. With lyophilized guinea-pig sera as a source of complement, similar dose–response curves were observed but a lower percentage of complement was bound per aggregate. Enzyme-linked immunosorbent assay Assay optimization. Initial cross titration assays determined that the optimal concentration of C1q for coating the plates was 0⋅5 mg (per well) while linear colour development was observed with a 1:10 000 dilution of 1 mg/ml of conjugate in PBS–Tween. Standard curve. Dose–response curves were constructed with varying amounts of Ig. The results shown in Figure 2 suggest that an Ig concentration of 0⋅5 mg/well may be the effective antigen concen100 90 80
Complement binding (%) Immunoglobulin in Aggregate concentration incubation mix (mg) 2⋅27% 4⋅35%
70 % Bound
Table 1. EP reference method: binding of aggregates to complement—variation with immunoglobulin concentration
60 50 40 30 20 10
0⋅5 1⋅0 2⋅5 5⋅0 10⋅0
8⋅5 8⋅5 12⋅2 12⋅2 16⋅8
23⋅2 25⋅8 31⋅7 38⋅6 32⋅1
20 CH50 units of complement were incubated with reference immunoglobulin preparations containing different aggregate concentrations.
0
1
2
3 4 % Aggregate
5 10
15
20
Figure 1. Reference method: dose–response curve for immunoglobulin aggregates. Aggregates formed by heating: reference Ig preparation at neutral pH were incubated with (W) 20 CH50 and (R) 5 CH50 units of complement. Intragam at (T) neutral pH and (Q) acid pH incubated with 20 CH50 units of complement.
BIO 25/1 ISSUE (827674)L—MS 0017
Ramasamy et al .
90
3
4 Absorbance (ELISA)
A
Absorbance
3
2
2
1
0
1 –1 1
2
3 % Aggregate
4
5
Figure 2. ELISA: standard curve constructed with different antigen concentrations. (Q) 1⋅92 (R) 1⋅0 (T) 0⋅5 (E) 0⋅2 (W) 0⋅1 mg Ig.
tration. Dose–response curves varied with aggregates formed under different conditions. Increasing concentrations of aggregates formed at acid pH caused only a slight increase of absorbance, suggesting poor binding to C1q-coated microtitre plates. The results were consistent with the low binding observed in Figure 1. A comparative standard curve with aggregates formed at neutral pH showed an increased binding with increasing aggregate concentrations (Fig. 3). At high levels of aggregate a plateau was reached. Correlation between the methods Complement binding as measured by a change in absorbance (ELISA) was compared to percentage binding by the reference method. There was 4
Absorbance
3
2
1
0
2.5
5.0
7.5 10.0 % Aggregate
12.5
15.0
0
17.5
Figure 3. ELISA: dose–response curve for immunoglobulin aggregates. Aggregates were formed by heating (Q) Reference Ig preparation at neutral pH, Intragam at (R) neutral pH (T) acid pH. 0⋅5 mg of Ig was used as the antigen.
100
25 50 75 % Bound (Ref method)
6 2.0 B Absorbance (ELISA)
0
1.5
1.0
0.5
0
10
20 30 40 50 60 % Bound (Ref method)
70
80
Figure 4. (A) Correlation of ELISA with reference method in aggregates formed from the reference immunoglobulin preparation at neutral pH. (B) Correlation of ELISA with reference method in 11 agggregates formed from heating: (Q) reference preparation at neutral pH; (R) intragam at neutral pH; (T) intragam at acid pH.
good correlation between the two assays with single aggregate types [ELISA = (0⋅03) ref. method −0⋅5; r = 0⋅98]. However, the correlation decreased [ELISA = (0⋅02) ref. method +0⋅003; r = 0⋅57] when the anticomplementary activity of aggregates formed at neutral and acid pH were compared (Fig. 4A,B). Assay reliability Within assay precision (CV) for the ELISA method with a sample containing a 4⋅4% aggregate content was 7⋅0% (n = 18). Between assay precision varied from 7⋅6% and 11⋅0% at aggregate concentrations of 4⋅4% (n = 8) and 2⋅3% (n = 11⋅0), respectively. For the reference method the percentage complement bound by the European Pharmacopoeia reference preparation varied from 10⋅5 to 13⋅6% and 78⋅9 to 100% when used as a negative and positive control, respectively, with fresh guinea-pig serum as the complement source. Control values were lower and more variable (3⋅3–24⋅9% and 46⋅2–68%) when
BIO 25/1 ISSUE (827674)L—MS 0017
Anticomplementary activity of intravenous immunoglobulins
commercial freeze-dried guinea-pig sera was used as the complement source. Discussion The current study shows that two major variable components in the reference method are the source and concentration of complement and the percentage aggregate concentration in the reaction mixture (Tables 1 and 2), thus confirming that ACA has to be measured under defined conditions, as specified by the reference method. Both the reference method and the ELISA show a dose response curve with the amount of complement bound increasing with the percentage aggregate content in immunoglobulin preparations (Figs 1 and 2). A comparative study of complement binding by immunoglobulin preparations with different aggregate types and concentrations showed that both methods correlated well (Fig. 4). Irrespective of the method used, an unacceptable level of ACA has to be defined. The current definition of an acceptable ACA is Q50% binding or 1 CH50/mg of protein. In analysing samples for ACA with the ELISA, a standard curve may be constructed with a set of standard Ig preparations of known complement binding capacity as determined with the reference method. This study indicates the need for such calibration material. Test samples considered positive will have an absorbance greater than the standard with 50% complement binding. Using this approach, the current study does not detect a significant difference in the ability to detect aggregates with unacceptable ACA between the two methods (Fig. 4). The ELISA measures the binding of aggregates to C1q and hence the potential for complement activation, through the classical pathway. Immunoglobulin aggregates can, although less efficiently, activate complement through the alternative pathway.11 The latter is more likely to be detected by the reference method. As a result, some variation between the two methods, due to differences in assay principles, is to be expected. However, the ELISA technique has the advantage in that it is easier to perform, more precise and less subject to reagent variability. It is not always possible to use fresh pooled sera from 10 guinea-pigs for the complement source as recommended by the reference method. The ELISA developed in the current study uses commercially available reagents easily accessible to the routine laboratory.
91
These results emphasize that the method of aggregate formation affects ACA and that estimation of the percentage distribution of aggregates by HPLC may not reflect ACA. Complement binding of aggregates, therefore, varies with their physical characteristics. The accepted upper limit for aggregate content in intravenous preparations is Q3%.10 The current study suggests that depending on the nature of aggregates, low levels (Q3%) can bind unacceptable amounts of complement (Fig. 1). In contrast aggregates such as those formed at acid pH, will bind complement poorly (Figs 1 and 3). These aggregates may cause side reactions by mechanisms other than their ACA, such as the downmodulation of the monocytes’ capacity to release oxygen metabolites and kill microbes.12 Both types of analysis, molecular weight distribution by HPLC and the measurement of ACA are required to fully characterize the final product. In summary, the present study shows that ACA is dependent on both the physical characteristic and concentration of aggregates in IVIG preparations. Therefore, a reliable technique to measure the ACA in such preparations becomes necessary. An ideal assay should be precise and allow for quantitative measurements of ACA. The results described here show good correlation between the two methods, though neither the reference method or the ELISA met all these requirements. However, on account of its simplicity the ELISA is the more suitable screening method. References 1. Hassig A. Intravenous Immunoglobulins: Pharmacological aspects and therapeutic use. Vox Sang 1986; 51: 10–17. 2. Tests for anticomplementary activity of immunoglobulins. European Pharmacopoeia Commission Guideline PA/PH/Exp.3T (95) 39, Def. 3. Kabat EA, Meyer MM. Experimental Immunochemistry; 2nd Edn. 1971. (Springfield, U.S.A., Thomas) pp. 133–154. 4. Prince AM, Horowitz B, Horowitz MS, Zang E. The development of virus free labile blood derivatives— a review. Eur J Epidemiol 1987; 3: 103–118. 5. European Union Working Party on Biotechnology/ Pharmacy—Guideline III/5544/94, 1994 6. Bjerrum L, Glikmann JC, Jensenium J, Svehag SE. Estimation of immune complexes by a microplate adapted C1q protein A enzyme linked immunosorbent assay. J Clin Lab Immunol 1983; 10: 53–57. 7. Jordan SC, Gautier E, Sakai R, Bahn L. Quantitation of circulating immunecomplexes in human serum by the Raji Cell and F(ab)2 anti-C3 microenzyme immunoassays. J Immunol Methods 1985; 83: 363–370.
BIO 25/1 ISSUE (827674)L—MS 0017
92
Ramasamy et al .
8. Yang YHJ, Ngo C, Ng yeh I, Uemura Y. Antibody Fc functional activity of intravenous immunoglobulin preparations treated with solvent detergent for virus inactivation. Vox Sang 1994; 67: 337–344. 9. Lowe RF, Galloway CJ, Dumas MC, Wong MF, Mitra G. Inactivation of hepatitis C virus in low pH intravenous immunoglobulin. Biologicals 1994; 22: 13–19. 10. Normal immunoglobulin for intravenous use: distribution of molecular size. British Pharmacopoeia 1994. pp. 1470.
11. Roitt I. Immunology. 4th Edn. 1996. pp. 13.1–13.5. 12. Eibl M, Ahmad R, Wolf HM, Linnau Y, Gotz E, Mannhalter JW. A component of Factor VIII preparations which can be separated from Factor VIII activity down modulates human monocyte function. Blood 1987; 69: 1153–1160.
Received for publication 19 June 1996; accepted 8 October 1996
Biologicals (1997) 25, 87–92
Measurement of Anticomplementary Activity in Therapeutic Intravenous Immunoglobulin Preparations Indra Ramasamy,* Em Tran and Albert Farrugia Molecular Biology, Therapeutic Goods Administration Laboratory, P.O. Box 100, Woden, Canberra, ACT 2606, Australia
Abstract. Anticomplementary activity (ACA) of aggregates in intravenous immunoglobulin preparations (IVIG) was investigated using the modified Kabat and Meyer classical complement consumption method recommended by the European Pharmacopoeia and a C1q-coated microtitre enzyme-linked immunosorbent assay (ELISA). The physical characteristics of aggregates were found to affect complement binding. Aggregates formed by heating IVIG preparations at acid pH bound complement poorly, while aggregates formed by heating IVIG at neutral pH showed high ACA. This suggests that analysis of complement binding capacity provides a level of aggregate characterization of aggregates which is additional to quantitation by High-performance liquid chromatography (HPLC). The correlation (r = 0·98) between the two tests was good when aggregates formed at neutral pH were compared, but decreased (r = 0·57) when aggregates formed at acid pH were included. A comparison of the results showed that there were no significant differences in the classification of aggregates with acceptable/unacceptable (i.e. pass/fail outcome) values of ACA. Between assay variation (CV = 7·6%) was lower in the ELISA test compared with the complement consumption assay (where percentage binding varied from 78·9% to 100%). Both assays are justified for the evaluation of ACA in therapeutic IVIG. The ELISA had the advantage in being more precise, less dependent on reagent source and requiring less technical expertise. = 1997 The International Association of Biological Standardization
Introduction Early attempts to administer immunoglobulin (Ig) preparations intravenously to patients with primary immunodeficiency were associated with severe anaphylaxis type reactions.1 The exact causes of such adverse reactions are not known, although a variety of theories have been put forward to explain them. The presence of aggregated IgG have been proposed as a cause of such reactions, particularly if these aggregates possess complement binding activity. The measurement of anticomplementary activity in immunoglobulin preparations is therefore an important aspect of their pharmaceutical quality control. A method based on complement mediated lysis of antibody-coated sheep red blood cells (SRBC) was adopted as the reference method for measuring *To whom correspondence should be addressed: Dr. I. Ramasamy, Molecular Biology, Therapeutic Goods Administration Laboratory, P.O. Box 100, Woden, ACT 2606, Canberra, Australia. 1045–1056/97/010087 + 6 $25.00/0/bg960063
ACA by the European Pharmacopeia (EP) in 1995.2,3 The method is cumbersome and not amenable to screening large numbers of samples. However, the need to measure ACA in Ig preparations has become more acute in recent years, due to the introduction of viral inactivation steps in the manufacture of plasma products. These include heat treatment at 60°C, which may generate aggregate formation and ACA. The tendency to form aggregates is further influenced by product formulation, and storage time and conditions. As manufacturers introduce viral inactivation steps under pressure from regulatory authorities and product consumers,4,5 the monitoring of the various manufacturing processes and each individual product for ACA becomes relevant. An alternative technique for measuring ACA uses an enzyme-linked immunosorbent assay principle. Aggregates in the Ig preparation are captured onto microtitre plate wells coated with C1q, the first step in the classical pathway for complement mediated lysis. Detection of bound aggregate is 7 1997 The International Association of Biological Standardization
BIO 25/1 ISSUE (827674)L—MS 0017
88
Ramasamy et al .
by horseradish peroxidase-conjugated protein A. The method’s initial optimization, using in house reagents, was for the measurement of circulating immunecomplexes in serum,6,7 but publications have shown the feasibility of using it to measure ACA in IVIG preparations.8 A problem with this approach is that investigators have used arbitrary units of measurement. However, the principle of biological standardization requires the availability of an appropriately calibrated standard, and the measurement of ACA in international units assigned through the reference method. This report is the first to compare the use of the EP and ELISA methods in measuring ACA in immunoglobulin preparations. Both methods were optimised for routine use, and aggregates formed under different conditions were then analysed.
Materials and methods Preparations The European Pharmacopoeia Ig preparation (50 g/l, pH 7⋅0), received as a lyophilized powder, from European Pharmacopoiea Commission (EPC) was used as both a negative control and at a higher concentration a positive control, as recommended by the EPC. Four batches of commercial IgG for intravenous use (Intragam, formulated at 60 g/l at pH 4⋅2) were obtained from CSL Bioplasma (Vic, Australia). The product is manufactured using Cohn fractionation followed by low pH incubation to suppress aggregate formation.9 Preparation of aggregated immunoglobulin Intragam was adjusted to pH 7⋅0 with 5 n NaOH, a concentrated solution of NaOH was used to minimize the change in volume. The reconstituted reference preparation, and Intragam at both acid (pH 4⋅2) and neutral pH (pH 7⋅0), were heated at 61°C in a water bath for various times and cooled to 4°C. The percentage aggregate formed was measured by HPLC gel filtration chromatography on a 0⋅6 m × 7⋅5 mm TSK gel G 3000SW column (ToSoh, Japan) as described by the European Pharmacopoiea.10 In the chromatogram the principal peak, molecular weight approximately 160 000, corresponded to the IgG monomer, and there were two further peaks corresponding to the dimer and aggregates. Low molecular weight degradation products were not observed. An increased rate of aggregate formation was found at acid pH compared with neutral pH.
Assay protocols 1. Complement binding analysis: Complement mediated lysis of antibody coated sheep red blood cells. The assay procedure was carried out according to the European Pharmacopoiea.2 Immunoglobulin preparation (10 mg) under investigation was incubated with guinea-pig complement in gelatin– barbitone buffer in 1 ml of reaction mixture. The titre of the remaining complement was measured by the lysis of rabbit antibody coated sheep red blood cells (Bio-Whittaker, USA). Anticomplementary activity of the immunoglobulin preparation is calculated as the percentage consumption of complement relative to the total complement in the reaction mix. Unless otherwise stated fresh guinea-pig complement was used; for comparison purposes lyophilized guinea-pig complement was obtained from two sources (BioMerieux, France and BioWhittaker). The haemolytic unit of complement activity (CH50 ) is the amount of complement that, under given reaction conditions will produce the lysis of 2⋅5 × 108 cells out of a total of 5 × 108 optimimally sensitized red blood cells. 2. Complement binding analysis: enzyme-linked immunosorbent assay. Flat bottomed microtitre plates (Immunlon, Dynatech) were coated overnight at 4°C with 100 ml of human C1q (Sigma Chemicals, St Louis, MO, USA) in 0⋅05 m bicarbonate/carbonate buffer pH 9⋅6. Plates were washed twice for 10 min in PBS–Tween and followed by a further wash for 1 h, the latter served as a blocking step. PBS–Tween was made up in 1 litre of distilled water containing 8⋅0 g NaCl, 0⋅2 g potassium dihydrogen phosphate, 1⋅15 g disodium hydrogen phosphate, 0⋅2 g potassium chloride, 1 g bovine serum albumin, 1 ml Tween-20, pH 7⋅4. Varying concentrations of test sample in 100 ml PBS were added to the microtitre plate and incubated for 3 h at 37°C. Following 3 further washes for 15 min, the plates were incubated for 1⋅5 h at 37°C with 100 ml of horseradish peroxidase-conjugated protein A (Sigma Chemicals) diluted in PBS–Tween. The plates were washed three times and 100 ml of substrate solution made up freshly with 40 mg o-phenylenediamine (OPD) and 40 ml of 30% hydrogen peroxide in 24⋅3 mls 0⋅1 m citric acid, 25⋅7 ml 0⋅2 m disodium hydrogen phosphate and 50 ml water, pH 5⋅0, was added. After incubation in the dark for 15 min at 37°C the reaction was stopped with 50 ml of 2⋅5 m sulphuric acid. The light absorption was read at 492 nm in a microplate reader (Titertrek,
BIO 25/1 ISSUE (827674)L—MS 0017
Anticomplementary activity of intravenous immunoglobulins
Flow laboratories, U.K.). When a standard curve was constructed curve fitting and analyte concentration interpolation were performed using a polynomial third order function.
Table 2. EP reference method: binding of aggregates to complement—variation with complement levels in incubation mix
RESULTS
Complement concentration (CH50 units)
Complement binding (CH 50/mg)
4⋅50 9⋅00 17⋅54
0⋅228 0⋅378 0⋅564
European Pharmacopoeia reference method Factors that affect complement binding by aggregates. Table 1 shows the variation of anticomplementary activity with amount of aggregate in the reaction mixture. Under defined conditions each individual aggregate preparation showed a characteristic maximum binding; an increased maximum binding of complement was found in preparations with higher concentrations of aggregate. This suggests that, as well as the amount of aggregate generated, the physical properties of the aggregates are influenced by the length of time the IVIG was heated, with a consequent effect on the maximum potential binding of each aggregate. In further experiments anticomplementary activity increased proportionately with complement levels in the incubation mix (Table 2). Standard curve. Figure 1 shows the relationship between aggregate concentration and complement binding of different preparations. At aggregate levels greater than 5%, complement binding plateaued; however, a linear relationship was observed at lower aggregate levels with aggregates formed at neutral pH. Aggregates formed at acid pH were incapable of binding complement at all aggregate levels tested, suggesting denaturation of the Fc region.
89
Reference immunoglobulin preparation (10 mg) with 4⋅35% aggregate concentration was incubated with varying amounts of complement. With lyophilized guinea-pig sera as a source of complement, similar dose–response curves were observed but a lower percentage of complement was bound per aggregate. Enzyme-linked immunosorbent assay Assay optimization. Initial cross titration assays determined that the optimal concentration of C1q for coating the plates was 0⋅5 mg (per well) while linear colour development was observed with a 1:10 000 dilution of 1 mg/ml of conjugate in PBS–Tween. Standard curve. Dose–response curves were constructed with varying amounts of Ig. The results shown in Figure 2 suggest that an Ig concentration of 0⋅5 mg/well may be the effective antigen concen100 90 80
Complement binding (%) Immunoglobulin in Aggregate concentration incubation mix (mg) 2⋅27% 4⋅35%
70 % Bound
Table 1. EP reference method: binding of aggregates to complement—variation with immunoglobulin concentration
60 50 40 30 20 10
0⋅5 1⋅0 2⋅5 5⋅0 10⋅0
8⋅5 8⋅5 12⋅2 12⋅2 16⋅8
23⋅2 25⋅8 31⋅7 38⋅6 32⋅1
20 CH50 units of complement were incubated with reference immunoglobulin preparations containing different aggregate concentrations.
0
1
2
3 4 % Aggregate
5 10
15
20
Figure 1. Reference method: dose–response curve for immunoglobulin aggregates. Aggregates formed by heating: reference Ig preparation at neutral pH were incubated with (W) 20 CH50 and (R) 5 CH50 units of complement. Intragam at (T) neutral pH and (Q) acid pH incubated with 20 CH50 units of complement.
BIO 25/1 ISSUE (827674)L—MS 0017
Ramasamy et al .
90
3
4 Absorbance (ELISA)
A
Absorbance
3
2
2
1
0
1 –1 1
2
3 % Aggregate
4
5
Figure 2. ELISA: standard curve constructed with different antigen concentrations. (Q) 1⋅92 (R) 1⋅0 (T) 0⋅5 (E) 0⋅2 (W) 0⋅1 mg Ig.
tration. Dose–response curves varied with aggregates formed under different conditions. Increasing concentrations of aggregates formed at acid pH caused only a slight increase of absorbance, suggesting poor binding to C1q-coated microtitre plates. The results were consistent with the low binding observed in Figure 1. A comparative standard curve with aggregates formed at neutral pH showed an increased binding with increasing aggregate concentrations (Fig. 3). At high levels of aggregate a plateau was reached. Correlation between the methods Complement binding as measured by a change in absorbance (ELISA) was compared to percentage binding by the reference method. There was 4
Absorbance
3
2
1
0
2.5
5.0
7.5 10.0 % Aggregate
12.5
15.0
0
17.5
Figure 3. ELISA: dose–response curve for immunoglobulin aggregates. Aggregates were formed by heating (Q) Reference Ig preparation at neutral pH, Intragam at (R) neutral pH (T) acid pH. 0⋅5 mg of Ig was used as the antigen.
100
25 50 75 % Bound (Ref method)
6 2.0 B Absorbance (ELISA)
0
1.5
1.0
0.5
0
10
20 30 40 50 60 % Bound (Ref method)
70
80
Figure 4. (A) Correlation of ELISA with reference method in aggregates formed from the reference immunoglobulin preparation at neutral pH. (B) Correlation of ELISA with reference method in 11 agggregates formed from heating: (Q) reference preparation at neutral pH; (R) intragam at neutral pH; (T) intragam at acid pH.
good correlation between the two assays with single aggregate types [ELISA = (0⋅03) ref. method −0⋅5; r = 0⋅98]. However, the correlation decreased [ELISA = (0⋅02) ref. method +0⋅003; r = 0⋅57] when the anticomplementary activity of aggregates formed at neutral and acid pH were compared (Fig. 4A,B). Assay reliability Within assay precision (CV) for the ELISA method with a sample containing a 4⋅4% aggregate content was 7⋅0% (n = 18). Between assay precision varied from 7⋅6% and 11⋅0% at aggregate concentrations of 4⋅4% (n = 8) and 2⋅3% (n = 11⋅0), respectively. For the reference method the percentage complement bound by the European Pharmacopoeia reference preparation varied from 10⋅5 to 13⋅6% and 78⋅9 to 100% when used as a negative and positive control, respectively, with fresh guinea-pig serum as the complement source. Control values were lower and more variable (3⋅3–24⋅9% and 46⋅2–68%) when
BIO 25/1 ISSUE (827674)L—MS 0017
Anticomplementary activity of intravenous immunoglobulins
commercial freeze-dried guinea-pig sera was used as the complement source. Discussion The current study shows that two major variable components in the reference method are the source and concentration of complement and the percentage aggregate concentration in the reaction mixture (Tables 1 and 2), thus confirming that ACA has to be measured under defined conditions, as specified by the reference method. Both the reference method and the ELISA show a dose response curve with the amount of complement bound increasing with the percentage aggregate content in immunoglobulin preparations (Figs 1 and 2). A comparative study of complement binding by immunoglobulin preparations with different aggregate types and concentrations showed that both methods correlated well (Fig. 4). Irrespective of the method used, an unacceptable level of ACA has to be defined. The current definition of an acceptable ACA is Q50% binding or 1 CH50/mg of protein. In analysing samples for ACA with the ELISA, a standard curve may be constructed with a set of standard Ig preparations of known complement binding capacity as determined with the reference method. This study indicates the need for such calibration material. Test samples considered positive will have an absorbance greater than the standard with 50% complement binding. Using this approach, the current study does not detect a significant difference in the ability to detect aggregates with unacceptable ACA between the two methods (Fig. 4). The ELISA measures the binding of aggregates to C1q and hence the potential for complement activation, through the classical pathway. Immunoglobulin aggregates can, although less efficiently, activate complement through the alternative pathway.11 The latter is more likely to be detected by the reference method. As a result, some variation between the two methods, due to differences in assay principles, is to be expected. However, the ELISA technique has the advantage in that it is easier to perform, more precise and less subject to reagent variability. It is not always possible to use fresh pooled sera from 10 guinea-pigs for the complement source as recommended by the reference method. The ELISA developed in the current study uses commercially available reagents easily accessible to the routine laboratory.
91
These results emphasize that the method of aggregate formation affects ACA and that estimation of the percentage distribution of aggregates by HPLC may not reflect ACA. Complement binding of aggregates, therefore, varies with their physical characteristics. The accepted upper limit for aggregate content in intravenous preparations is Q3%.10 The current study suggests that depending on the nature of aggregates, low levels (Q3%) can bind unacceptable amounts of complement (Fig. 1). In contrast aggregates such as those formed at acid pH, will bind complement poorly (Figs 1 and 3). These aggregates may cause side reactions by mechanisms other than their ACA, such as the downmodulation of the monocytes’ capacity to release oxygen metabolites and kill microbes.12 Both types of analysis, molecular weight distribution by HPLC and the measurement of ACA are required to fully characterize the final product. In summary, the present study shows that ACA is dependent on both the physical characteristic and concentration of aggregates in IVIG preparations. Therefore, a reliable technique to measure the ACA in such preparations becomes necessary. An ideal assay should be precise and allow for quantitative measurements of ACA. The results described here show good correlation between the two methods, though neither the reference method or the ELISA met all these requirements. However, on account of its simplicity the ELISA is the more suitable screening method. References 1. Hassig A. Intravenous Immunoglobulins: Pharmacological aspects and therapeutic use. Vox Sang 1986; 51: 10–17. 2. Tests for anticomplementary activity of immunoglobulins. European Pharmacopoeia Commission Guideline PA/PH/Exp.3T (95) 39, Def. 3. Kabat EA, Meyer MM. Experimental Immunochemistry; 2nd Edn. 1971. (Springfield, U.S.A., Thomas) pp. 133–154. 4. Prince AM, Horowitz B, Horowitz MS, Zang E. The development of virus free labile blood derivatives— a review. Eur J Epidemiol 1987; 3: 103–118. 5. European Union Working Party on Biotechnology/ Pharmacy—Guideline III/5544/94, 1994 6. Bjerrum L, Glikmann JC, Jensenium J, Svehag SE. Estimation of immune complexes by a microplate adapted C1q protein A enzyme linked immunosorbent assay. J Clin Lab Immunol 1983; 10: 53–57. 7. Jordan SC, Gautier E, Sakai R, Bahn L. Quantitation of circulating immunecomplexes in human serum by the Raji Cell and F(ab)2 anti-C3 microenzyme immunoassays. J Immunol Methods 1985; 83: 363–370.
BIO 25/1 ISSUE (827674)L—MS 0017
92
Ramasamy et al .
8. Yang YHJ, Ngo C, Ng yeh I, Uemura Y. Antibody Fc functional activity of intravenous immunoglobulin preparations treated with solvent detergent for virus inactivation. Vox Sang 1994; 67: 337–344. 9. Lowe RF, Galloway CJ, Dumas MC, Wong MF, Mitra G. Inactivation of hepatitis C virus in low pH intravenous immunoglobulin. Biologicals 1994; 22: 13–19. 10. Normal immunoglobulin for intravenous use: distribution of molecular size. British Pharmacopoeia 1994. pp. 1470.
11. Roitt I. Immunology. 4th Edn. 1996. pp. 13.1–13.5. 12. Eibl M, Ahmad R, Wolf HM, Linnau Y, Gotz E, Mannhalter JW. A component of Factor VIII preparations which can be separated from Factor VIII activity down modulates human monocyte function. Blood 1987; 69: 1153–1160.
Received for publication 19 June 1996; accepted 8 October 1996