Influence of plasma immunoglobulin level on antibody synthesis

The regulation of serum antibody levels has been a subject of debate for several decades. A large number of studies have provided evidence for enhanced antibody synthesis after immunoglobulin depletion and thus for the existence of an immunoregulatory feedback mechanism.1-7 In contrast, Charlton and colleagues8-10 demonstrated with a series of animal experiments that the rapid increase of antibody serum levels after plasmapheresis could be explained by immunoglobulin backflow from extravascular space and decreased catabolism alone. More than 10 years later the lack of an increased antibody synthesis in a low immunoglobulin state was confirmed in the excellent studies by Junghans and Anderson11 and by Junghans.12 Knock-out mice with disrupted immunoglobulin G (IgG) protection receptors, which effect increased catabolism of IgG and subsequently low IgG serum levels, had a similar biosynthesis rate as normal control subjects.

The complex metabolism of IgG has so far prevented exact measurement of immunoglobulin synthesis in humans with low antibody levels. However, a prerequisite for immunoglobulin assembly is the synthesis of free light chains (flcs) that are produced in excess during antibody production and secreted as free κ or λ monomers into the vascular space.13,14 Because flcs have a half-live of only 2 to 4 hours,15 they reflect current immunoglobulin production and can be used for estimating the activity of diseases accompanied by autoantibody synthesis.16,17 We hypothesized that increased antibody synthesis after immunoglobulin depletion would be accompanied by increased flc serum levels and measured free κ and λ light chain concentrations in 8 patients undergoing immunoadsorption (IA) therapy.

The study protocol was approved by the local ethics committee, and the patients' informed consent was obtained. Eight patients with various autoimmune diseases and normal kidney function treated with IA therapy were recruited (Table 1). Concomitant immunosuppressive therapy was prednisolone (n = 4), mycophenolate mofetil (n = 3), and cyclosporine A (n = 1); 3 patients received no additional therapy. We recruited additionally 3 patients with familial hypercholesterolemia, who were treated with regular low-density lipoprotein (LDL)–cholesterol apheresis, as control subjects. The IA and LDL apheresis therapy procedure used at our institution was described previously.18,19 In brief, peripheral venous blood was drawn and the plasma separated by centrifugation. Thereafter the plasma was treated with either Ig-Therasorb columns (Therasorb, Munich, Germany), coated with sheep antihuman immunoglobulin antibodies, or LDL-Therasorb columns (Therasorb), coated with anti-LDL antibodies. Blood samples were analyzed immediately before and after plasmapheresis and at days 1, 3, and 5 after treatment. Measurement of flc was performed by a latex-enhanced immunoassay20 with a lower detection range for free κ and λ light chains of 3.4 mg/L. The blood levels of serum immunoglobulins (IgG, IgA, IgM) and of specific autoantibodies (thyroglobulin, acetylcholine receptor, double-stranded DNA (dsDNA), and antinuclear antibody/antinuclear factor) were determined with the use of standard methods. To exclude only B-cell activation by IA therapy or activation of the underlying disease, which would have increased immunoglobulin synthesis and interfered with the flc synthesis rate, in vivo B-cell activity was measured by 4-color flow cytometry before and during the post-IA period by using 3 different antibodies (CD38, CD43, and CD22; all antibodies by Immunotech, Marseille, France). Activation of B cells was defined as an increase of CD38 and CD43 expression and reduction of CD22 expression.

The results are expressed as median and range. Statistical analysis was carried out by the Student t test for paired or unpaired data or Wilcoxon signed rank test. A P value < .05 was considered statistically significant.

Free κ light chain levels decreased in 7 of 8 patients during IA and remained unchanged in one patient (Table2). The levels increased to pretreatment values in most patients within 1 day (median difference to pretreatment values, −2% [range, −41% to 62%] with one patient below detectable limit) and remained stable thereafter (Table 2). Only one patient had a κ light chain pretreatment level above normal range (patient 8), which decreased by IA but remained slightly above normal range during the observation period (normal, 8.4 ng/mL; range, 3.6-15.9 ng/mL). Flc of the λ type decreased similarly during IA in all patients (Table 2). Almost all patients reached pretreatment values within 24 hours (median difference, −1%; range, −14% to 15%). The levels remained within normal range during the further course (normal, 14.5 ng/mL [range, 8.1-33 ng/mL]). The median κ/λ ratio in patients with measurable free κ light chain levels was 0.49 (range, 0.33-1.30) before treatment and remained within normal range during the posttreatment period (mean, 0.54 [range, 45-0.63]; normal, 0.6 [range, 0.36-1.0]). Free κ and λ light chain levels before and after therapy in the 3 control patients with LDL apheresis remained unchanged throughout the study (mean κ, 15 ng/mL [range, 13-17 ng/mL] and mean λ, 18 ng/mL [range, 18-19 ng/mL]). IgG serum levels decreased by approximately 90% during IA and increased to approximately 70% of pretreatment levels within 5 days (Table3). There was a tendency toward a slower increase of serum levels in patients receiving no adjunctive immunosuppressants (n = 4) than in patients with additional mycophenolate mofetil, cyclosporine A, or prednisolone (> 10 mg/day) therapy (median, 69%; range, 63%-85% versus 81%; range, 63%-94%;P = NS). IgM and IgA decreased by 61% (range, 47%-81%) and 68% (range, 47%-81%), respectively, and increased within 5 days to 75% (range, 52%-111%) and 81% (range, 79%-98%), respectively, of pretreatment levels. Immunoglobulin levels remained unchanged in the control patients treated with LDL apheresis. Specific antibody levels decreased substantially during IA in all patients with identifiable autoimmune disease and increased to pretreatment values in 4 of 7 patients within 5 days (Table 3).

The 3 different activity markers (CD22, CD38, and CD43) identified on CD19⁺ lymphocytes showed a normal distribution in 6 of 8 investigated patients and were not influenced by IA (data not shown). Two patients had activated B cells prior to initiation of therapy. Patient 6 had markedly increased CD38⁺ B cells (98%), which decreased after IA (78%). In patient 8, CD38⁺ (78%) and CD43⁺ (74%) were increased and CD22⁺(42%) decreased and normalized after IA therapy (CD38⁺, 19%; CD43⁺, 21%; and CD22⁺, 94%). The 3 patients treated with LDL apheresis had a normal distribution of CD38⁺, CD43⁺, and CD22⁺ B cells throughout the study.

Thus, we were able to demonstrate that immunoglobulin depletion with IA does not influence antibody synthesis which agrees with previous animal experiments describing the lack of an immunoglobulin feedback mechanism.8-11 The close correlation between flc concentration and immunoglobulin synthesis has been demonstrated in several conditions associated with B-cell activation, such as multiple sclerosis16 and systemic lupus erythematodes.17 Because B-cell activation and antibody synthesis precedes tissue injury by several weeks, flcs might even be used as reliable predictors for subsequent clinical relapse of disease.21 Thus, by demonstrating that flc serum levels remained unchanged after immunoglobulin depletion, we were able to prove that a reduction of antibody serum levels is not followed by an increased biosynthesis. Moreover, because immunoadsorption therapy not only reduced IgG but also IgM and IgA serum levels by 60% to 70% because of the use of a nonspecific adsorption column, an intrinsic feedback mechanism of other immunoglobulin classes might be excluded as well.

The lack of an accelerated antibody synthesis after immunoglobulin depletion makes the widely accepted strategy of administering cytotoxic drugs during plasmapheresis, to prevent an accelerated antibody production,22 obsolete. The insufficiency of prophylactic immunosuppressive therapy on antibody rebound was demonstrated in our patients by the similar immunoglobulin increase in those treated with additional immunosuppressants, when compared with patients without concomitant cytostatic drugs. Antibody levels as well as flc levels did not differ between the groups and increased to pretreatment values equally fast. Similar to others9,12 we cannot, therefore, recommend the administration of immunosuppressants for the prevention of an increased antibody synthesis after plasmapheresis. However, patients with a chronic autoimmune disease, treated with plasmapheresis either as a result of acute exacerbation or as a long-term therapeutic strategy, will, nevertheless, profit in most cases from the long-term effects of immunosuppressants.

The absence of an increase of B-cell activity after IA contradicts a previous study7 but, nevertheless, excludes the possibility that IA therapy itself might promote antibody synthesis by release of antigenic peptides. On the contrary, as indicated in the patient with active Guillain-Barré syndrome, characterized serologically by highly activated B cells and increased κ light chains, removal of autoantibodies and immune complexes might lead to a normalization of both B-cell activity markers and flc levels. It is, therefore, tempting to speculate that antibody removal might not only be without influence on immunoglobulin synthesis but also by interrupting the inflammatory process even decrease B-cell activity and antibody production.

In conclusion, our study shows that antibody synthesis in humans is not regulated by a biofeedback mechanism. These findings are in line with similar results found in rabbits and knockout mice.8-11Administration of cytotoxic drugs for inhibition of a supposed antibody rebound should, therefore, be avoided, because it has no influence on postapheresis synthesis rate but implies an additional risk for the patient.

The flc Kits were provided by “The Binding Site,” Birmingham, England.

We thank Professor I.C.M. Maclennan, Birmingham, United Kingdom, and Professor G. Zlabinger, Vienna, Austria, for valuable discussion, and Mrs. Konstantin and Mrs. Czarnecki for excellent technical assistance.