Immunoadsorption as a Tool for the Immunomodulation ofthe Humoral and Cellular Immune System in

Autoimmune Disease

Norbert Braun and Teut Risler

Sektion Nieren- und Hochdruckkrankheiten, Universitatsklinikum Tubingen, Tubingen, Germany

Abstract: Immunoadsorption onto staphylococcal proteinA is a newly developed semiselective extracorporeal ad-sorption technique for immunoglobulins applied in pa-tients suffering from severe autoimmune disease. Its effecton the humoral and cellular immune system was investi-gated using standard immunological assays. The elimina-tion capacity for total IgG and IgG subclasses 1, 2, and 4was more than 90% but for subclass IgG3 varied between30 and 90%. Autoantibodies, e.g., anti-dsDNA, anti-glomerular basement membrane (anti-GBM), anti-cardiolipin, and anti-human leukocyte antigen (anti-HLA)antibodies, were eliminated in comparable amounts. Theaffinity of protein A for circulating immune complexes

(CIC) was 300 times greater than for soluble IgG. HLA-IIexpression on monocytes and T lymphocytes was reducedover time during repeated IAs (IA). The number of acti-vated T lymphocytes declined while the percentage of na-ive T cells increased. A diminished CD4/CD8 ratio nor-malized during IA treatment. These results indicate thatIA actively modulates the humoral as well as the cellularimmune system in addition to its immunoglobulin reducingeffect. Key Words: Immunoadsorption—Staphylococcalprotein A—Autoimmune disease—Systemic lupus erythe-matosus—Immune system—Circulating immune com-plexes—T lymphocytes—Immunoglobulins.

Extracorporeal elimination techniques for the re-moval of pathological substances from blood andplasma of patients have been known for over 30years. However, conventional plasma exchangetherapy does not only remove harmful agents butalso natural plasma components (e.g., blood coagu-lation factors and hormones). Although plasma ex-change and plasmapheresis have been widely used inmany clinical indications, their therapeutic effectsrarely have been proven by controlled trials. Thus,research has focused on other more selective elimi-nation techniques. Plasma cryofiltration and semi-selective adsorption onto biological and artificialsurfaces have been applied in in vitro and in clinicalsettings. In vivo immunoadsorption (IA) onto char-coal has been shown to remove ssDNA (1). Dextransulfate (2), phenylalanine, and tryptophan (3) havebeen found to bind fibrinogen and immunoglobulins

(4). Immunoglobulins and circulating immune com-plexes (CIC) bind to staphylococcal wall protein A(5). Recently, immunological devices have been de-veloped for the highly selective removal of immuno-globulins (6) and C1q fixing immune complexes (7).Now, at the end of the present millennium, severaldifferent adsorption techniques are available for thetreatment of autoimmune disorders. Although theireffectiveness in removing antibodies and other sub-stances has been demonstrated in the past, the exactmode of action on the host immune system has notbeen a major point of research interest. This papersummarizes the results of clinical and immunologicalinvestigations on the effect of protein A IA in pa-tients suffering from autoimmune disease.

CASE REPORT

A 17-year-old Caucasian girl with membranopro-liferative lupus nephritis (World Health Organiza-tion [WHO] class IV) first diagnosed in 1987 wasasymptomatic until September 1993, receiving aza-thioprine and corticosteroids. She was admitted to

Received January 1999.Address correspondence and reprint requests to Dr. Norbert

Braun, Medizinische Universitatsklinik Tubingen, SektionNieren- und Hochdruckkrankheiten, Otfried-Muller-Str. 10,72076 Tubingen, Germany. Email: nbraun@uni-tuebingen.de

Therapeutic Apheresis3(3):240–245, Blackwell Science, Inc.© 1999 International Society for Apheresis

240

our unit because of increasing ankle edema. A renalbiopsy showed diffuse mesangioproliferative glo-merulonephritis with focal glomerular scarring(WHO class IVd). Treatment was started using theprotocol of the Lupus Plasmapheresis Study Group(8) with i.v. cyclophosphamide. The patient was dis-charged but had to be readmitted as an emergencythe following month because of weight gain of morethan 6 kg/week and deterioration of her renal func-tion (serum creatinine, 1.5 mg/dl and proteinuria, 12g/24 h). She was treated by 3 consecutive plasma-phereses followed by i.v. cyclophosphamide and oralcorticosteroids (8). The condition of the patient im-proved temporarily, proteinuria declined but still re-mained in the nephrotic range, and the patient couldleave the hospital.

The following month she was admitted again toour emergency department with fever >40°C, jointpain, and multiple lymphoma on her neck. A secondplasmapheresis treatment with subsequent i.v. cyclo-phosphamide was administered. Renal function de-teriorated gradually, and the patient became dialysisdependent. Despite adequate fluid removal, bilateralpulmonary infiltrates, pleura, and pericardial effu-sions developed. Scintillating scotoma and hemiano-pia were noticed despite high dose prednisolonetherapy.

The patient became unresponsive. A few dayslater, she had to be admitted to the intensive careunit with recurrent grand mal seizures. With agree-ment from her parents, plasmapheresis, predniso-lone, and cyclophosphamide treatment was stopped.A week later, treatment was started with IA ontoprotein A without additional immunosuppressivetreatment.

The patient’s symptoms resolved within a week(Fig. 1). Two weeks after initiation of IA, intermit-tent dialysis was stopped, and the patient was startedon oral cyclophosphamide in the third week. IA wascarried out for 6 weeks 3 times a week to avoid therebound phenomenon. A repeated biopsy was per-formed 9 months later and primarily showed onlyminor signs of activity and chronicity. At that timethe cyclophosphamide was replaced by azathioprineand prednisolone 10 mg every other day. Her anti-dsDNA Ab titer was 25 IU/ml (Crithidia antigennegative) at the last follow-up. Serum creatinine was1.1 mg/dl, and proteinuria ranged between 3 and 4g/24 h.

IMMUNOADSORPTION TECHNIQUE

As a result of the encouraging positive results ofIA in the described patient, a treatment protocol was

developed which consisted of stopping all immuno-suppressive treatments at least 2 weeks before IA.Prednisolone therapy was tapered down to a maxi-mum of 10 mg/day when prescribed for more than 6months. Written informed consent was obtainedfrom all patients or their next of kin. Immunoadsorp-tion was begun every day with a minimum of 1 rela-tive body plasma volume (BPV) (9), and the BPVwas adjusted according to final total immunoglobu-lin levels with the goal of a reduction of at least 20%of the baseline levels. In most cases, between 2 and3 BPV were treated per session. Blood was drawnfrom the patient using either a peripheral vein ormost often a central vein catheter. Anticoagulationwas achieved with a 2,500 IE sodium heparin bolusand a continuous infusion of 1,000 IU sodium hepa-rin/h in addition to acid citrate dextrose-A (ACD-A)at a ratio of 1:12 (citrate/whole blood). Heparin wasavoided in patients suffering from Goodpasture’ssyndrome because severe lung bleeding had beenpreviously observed. In patients with myastheniagravis, the ACD-A ratio was reduced to 1:30 becauselow plasma calcium levels aggravated general muscleweakness. Plasma was usually obtained by centrifu-gation, but occasionally membrane plasma separa-tion was used.

The plasma flow rate was steadily increased from10 to 35 ml/min and processed through a twin pro-tein A column pair (Immunosorba, Excorim, Lund,Sweden) using the Citem 10 plasma monitor (Ex-corim). Both columns were alternately loaded withplasma and rinsed with citrate buffer using cyclesbetween 7 and 10 min, dependent on the immuno-globulin concentration in the processed plasma indi-rectly determined by UV adsorption. Intravenousimmunoglobulin substitution was avoided wheneverpossible. Periods between IA sessions were pro-longed when remission was achieved, and overlap-ping oral immunosuppression with cyclophospha-mide was begun to avoid antibody rebound.

HUMORAL IMMUNE SYSTEM

Immunoglobulins, immune complexes, andcomplement factors are involved in the evolution ofautoimmune diseases such as systemic lupus erythe-matosus, myasthenia gravis, etc. The staphylococcalwall protein A binds to the Fc part of human IgG1,IgG2, and IgG4 and to the Fab part of IgG3, IgM,and IgA in a nonimmunological manner (KD 4 2.2× 10−8 M) (10). Binding to the Fab part of immuno-globulins (alternative binding site) is dependent onthe utilization of the VHIII-gene for the heavy Igchain (11).

IMMUNOADSORPTION 241

Therapeutic Apheresis, Vol. 3, No. 3, 1999

Investigations of the kinetics of antibody removalduring IA were either performed in vitro or in pa-tients undergoing IA and concomitant immunosup-pression. Neither approach allows for identificationof the exact role of IA in the reduction of autoanti-body levels. We tested the effect of IA onto proteinA on antibody and autoantibody levels in patientswithout oral or i.v. immunosuppression. Antibodylevels were determined before, during, and aftertherapy in a total of 86 individual IA sessions. One-and two-compartment models were applied for thedescription of total IgG, IgG subclasses, anti-humanleukocyte antigen (anti-HLA) antibody titers, anti-glomerular basement membrane (anti-GBM) IgG,anticardiolipin, anti-dsDNA IgG, and anti-dsDNAIgG3. The two-compartment model fit best in mostcases as determined by Akaike’s information crite-rion. The elimination half-time was estimated using asingle and a two-compartment model analysis. Theelimination half-time for total IgG was estimated as3.6 days for alternate day treatment. Daily treatmentreduced the elimination half-time to 2.9 days. Ki-netic data for IgG subclasses are presented in Table1. In selected cases a shorter half-life for IgG3 wasobserved compared to that of total IgG, which canbe explained by the shorter, natural generation rateof IgG3 and possible smaller distribution space. Onthe other hand, IgG3 levels were not completely sup-pressed by IA onto protein A in all patients as com-pared with IgG subclasses 1, 2, and 4. Other adsorp-tion devices, based on immunological binding ofhuman immunoglobulins (i.e., Ig-Therasorb, Thera-sorb, Unterschlessheim, Germany), have beenshown to reduce IgG3 levels below detection limitregardless of their Fab proprieties. At present, it is

FIG. 1. The fundoscopic images from a patient suffering fromsevere SLE demonstrate retinal hemorrhage (a, big arrows), ir-regularities of retinal arteries (a, small arrows), and a blurredoptic disk (a). After treatment with immunoadsorption onto pro-tein A for 3 weeks, the hemorrhages subsided, and only minorirregularities of retinal arteries were seen (b, arrows). Aftercompletion of a 6 week course of immunoadsorption followed byoral cyclophosphamide, the patient exhibited an almost normalfundoscopy (c).

TABLE 1. Kinetic data for IgG subclasses

f(IgG1) f(IgG2) f(IgG3) f(IgG4)

n 17 16 40 17Minimum 0.387 0.218 −0.708 0.24625% percentile 0.744 0.446 −0.013 0.330Median 0.989 0.620 0.1 0.65075% percentile 1.175 0.930 0.347 0.942Maximum 2.561 1.304 1.084 1.697Mean 1.198 0.678 0.157 0.76StdDev 0.673 0.323 0.342 0.452StdError 0.163 0.080 0.054 0.109Lower 95% Cl 0.851 0.506 0.048 0.528Upper 95% Cl 1.544 0.850 0.267 0.993

Adsorption efficacy was calculated by f(t) 4 [-log(Ct/C0)](PVt/Vtotal) (18), where C is the concentration during immunoadsorp-tion, C0 is the initial concentration, the quotient PV/Vtotal isequivalent to BPV, where PV represents the processed plasmavolume, and Vtotal the hematocrit corrected circulating plasmavolume of the patient at any time of the adsorption procedure. Forcomparison, f was calculated for the complete duration of anyindividual immunoadsorption.

N. BRAUN AND T. RISLER242

Therapeutic Apheresis, Vol. 3, No. 3, 1999

not known whether these new devices bind all allo-types of antibodies because the development of theadsorption columns was technically dependent on animmunoglobulin pool primarily collected fromNorth American donors.

After the cessation of IA, a rapid and steady in-crease of antibody levels was noted in all patients.This rebound is due to the redistribution of antibod-ies from the interstitial space into the vasculature.There was a wide range of redistribution half-timescovering 1–18 days. The redistribution half-time cor-relates with the amount of processed plasma and theduration of IA (kre 4 0.01–0.007 h−1). Effective IAfor at least 14 days beginning with daily treatmentand alternate treatment in the second week usuallyresults in a very slow increase of antibody levels re-flecting the natural resynthesis rate of immunoglob-ulins. Additive immunosuppression will suppress thefurther increase of antibody levels in this state andresult in a significant longer resynthesis half-time.New autoantibody formation can be prevented bythis procedure.

The elimination of autoantibodies, e.g., anti-dsDNA, anticardiolipin, and anti-GBM (Fig. 2) an-tibodies, correlated well with the elimination of totalIgG and IgG1. Interestingly, anti-dsDNA IgG3 wasalso reduced in all patients except 1 who receivedadditional daily i.v. immunoglobulins for septicemia.In this patient, an increase of anti-dsDNA IgG3 wasobserved. Anti-HLA antibodies were tested usingfluorescence activated cell sorter (FACS) cross-match and conventional crossmatches. While totalIgG was eliminated according to our experience inother patients with autoimmune diseases, anti-HLAantibody titers behaved differently. Most of the titerswere reduced, but reactivity to some distinct HLA

antigens was even increased during IA (Fig. 3). Itwas suggested that low titer antiidiotypic IgG anti-bodies were preferably bound to protein A, releasingmore free anti-HLA antibodies into the circulationof the patient (12). Pretransplant antibody screeningwith in vitro and in vivo test IA seems to be man-datory for successful transplantation of highly HLAsensitized patients.

CIC were detected in several patients with severesystemic lupus erythematosus (SLE) by ELISAtechnique sensitive for either C1q fixing or C3 fixingCIC. Although crossreaction with C1q antibodiescannot be ruled out at present, it was noted thatthese tests became readily negative within a singleIA procedure (13). The rapid elimination of CIC isconsistent with a high affinity of staphylococcal pro-tein A for immune complexes, which is 300 timesgreater than that for soluble IgG (5).

While it is well known that extracorporeal treat-ment leads to an activation of the complement sys-tem via activation of the alternative pathway due tocontact with artificial membranes and surfaces, pa-tients with active SLE have very low total C3 and C4levels. Immunoadsorption is no exception to thisphenomenon and necessitates i.v. anticoagulationwith citrate (ACD-A) (14) to suppress the genera-tion of C3a. Investigations using different ACD-Aratios showed that a relation of ACD-A/whole bloodof more than 1:12 is adequate. C3 and C4 levels thenincrease during adequate IA treatment in SLE pa-tients and can be used for immunological monitoringas long as no i.v. immunoglobulins have been substi-tuted. On the other hand, i.v. immunoglobulin treat-ment for SLE, although used as an alternative treat-ment if immunosuppression is contraindicated orharmful, can lead to excessive formation of CIC anda rapid decline of C3 and C4.

FIG. 2. The graph represents the elimination of anti-GBM anti-bodies in a patient with Goodpasture’s disease and severe lunginvolvement. The upper line denotes anti-GBM antibody titersbefore and the lower line after immunoadsorption. The area be-tween the 2 dashed lines represents antibody titers of unknownsignificance. Tests were performed by Prof. Dr. M. Weber, Mediz-inische Klinik I, Krankenhaus Mehrheim, Koln, Germany.

FIG. 3. In terms of single fluorescence immunostaining of mono-cytes from a patient with SLE undergoing immunoadsorptiononto protein A without concomitant drug immunosuppression,the grey histogram shows HLA-II expression on monocytesafter completion of 4 weeks of immunoadsorption while theopen histogram represents HLA-II+ monocytes before immuno-adsorption.

IMMUNOADSORPTION 243

Therapeutic Apheresis, Vol. 3, No. 3, 1999

CELLULAR IMMUNE SYSTEM

While IA was originally designed to remove anti-bodies from the circulation, it is currently knownthat extracorporeal treatment influences the cellularimmune system, primarily in a negative (suppres-sive) way. It has been reported that hemodialysisinfluences the T-cell subpopulation and activates themonocyte/macrophage system, dependent on the ap-plied dialyzer material (15). Premature results of in-vestigations in our group addressing the cellular im-mune system during extracorporeal IA showed atransient increase in activated monocytes. In patientsundergoing membrane plasma separation instead ofcentrifugation, this rise in HLA-II expressing mono-cytes was more prominent. Surprisingly, the num-bers of HLA-II+ monocytes declined when IA wasrepeated over a longer period of time (Fig. 4).

Changes in the T-cell population were more com-plex. Patients with severe SLE showed a depressedCD4/CD8 ratio (mean, 0.8). Prolonged intermittentIA resulted in a normalization of this ratio whenremission was achieved. In very few patients not suf-fering from severe SLE and with initially normalCD4 counts, a transient decrease of CD4+ cells was

noted. No substantial changes of expression of IL-2receptors (CD25) were found in these patients whileCD25+ cells were rarely found in patients with primarylow CD4 counts and remission due to IA. Antigen-activated T cells (CD3+CD45RO+) were regularlyseen before IA was begun (mean before IA, 58%).

Immunoadsorption seemed to reduce the numbersof these cells irrespective of the CD4 or CD8 phe-notype (mean after last IA, 53%) and increased theimmature T-cell fraction (CD3+CD45RA+; 21% be-fore IA and 42% after last IA; Fig. 5). Despite thepreliminary character of these experiments, the re-sults can be compared to historical findings in pa-tients with SLE undergoing intensive plasma ex-change therapy where an increase in the rosetteformation of T cells and an improvement of concon-valin A mediated suppressor activity was reported inaddition to a decrease of antilymphocyte antibodies(16,17).

IMMUNOADSORPTION AS A TOOL FORACTIVE IMMUNOMODULATION?

The application of extracorporeal IA enables thephysician for the first time to remove pathogenic

FIG. 4. The scatter plot of triplefluorescence staining for livingCD45RA expressing T lympho-cytes is of the same patient asshown in Fig. 3. A shift fromCD3+/CD45RA− (a, left upperquadrant, pre IA) to CD3+/CD45RA+ cells (b, right upperquadrant, post IA) was seen dur-ing prolonged and repeated immu-noadsorption in this patient suffer-ing from severe SLE. The totalnumber of T lymphocytes was in-creased as seen by the lower num-ber of dots seen in the left lowerquadrant (b).

FIG. 5. The FCXM are of 2 pa-tients undergoing pretransplantimmunoadsorption onto protein Ato reduce anti-HLA antibody ti-ters. In patient HL, serum wastested against cells with HLA typeA24,26 (a) and A2,33 (b), and thepatient’s TU serum was testedagainst cells with HLA-B7 (c).Gray histograms denote autolo-gous control, Pre-IA: before im-munoadsorption, Post-IA: afterimmunoadsorption. It is shownthat reactivity against HLA-A24,26 is markedly reduced (a)while HLA reactivity is increased(b) and (c).

N. BRAUN AND T. RISLER244

Therapeutic Apheresis, Vol. 3, No. 3, 1999

antibodies and most likely CIC in substantialamounts from a patient’s circulation. The major ad-vantage of this technique is its simplicity and safety,which allows adequate and repeated treatment for along period of time. The clinical success correlateswith changes in immunoglobulin and complementlevels but also with more sophisticated markers suchas immune complexes and surface markers of lym-phocytes and monocytes. Because all results de-scribed above were obtained from patients receivingno concomitant immunosuppression at the time ofIA, these changes in humoral and cellular immunemarkers must be attributed to either spontaneousdisease related phenomena or more likely to thetreatment with IA onto protein A. The immunologi-cal effects of IA either can be attributed to theprimary plasma separation to obtain the immuno-globulin and CIC containing plasma for further pro-cessing or to the elimination of immunoglobulinsand CIC via binding to protein A itself. One canassume that the high affinity of protein A for CICresults in an improved clearance of soluble immunecomplexes preventing further vasculitic lesions in pa-tients with severe disease. While this immediate ef-fect might be responsible for the rapid improvementin patients with impaired kidney function in immunecomplex mediated lupus nephritis, changes in T-cellpopulations require prolonged and repeated treat-ment.

We therefore assume that CIC contain disease rel-evant antigens which usually are presented to T cellsvia HLA-II+ macrophages. Removal of autoantigencontaining CIC should then result in a lack of selfantigen for the presentation to autoreactive T cells.HLA-II expression on monocytes will be downregu-lated, and the relationship between antigen-acti-vated CD45RO+ and immature CD45RA+ T cellswill be shifted toward the immature cell fraction.The levels of autoantibodies, i.e., anti-dsDNA anti-bodies, will decrease as a consequence of diminishedautoreactive T-cell activity. This hypothesis is cur-rently supported by the rapid induction of remissionin patients with severe autoimmune diseases, the de-crease of activated monocytes, the delayed rise ofthe immature T cell subpopulation, and finally by theprolonged suppression of autoantibody formationafter cessation of IA. Our present objective is toidentify disease relevant antigens in the eluates ofprotein A used in extracorporeal IA of patients withautoimmune diseases.

REFERENCES

1. Terman DS, Buffaloe G, Mattioli C, Cook G, Tillquist R,Sullivan M, Ayus JC. Extracorporeal immunoadsorption: Ini-tial experience in human systemic lupus erythematosus. Lan-cet 1979;2:824–7.

2. Kutsuki H, Takata S, Yamamoto K, Tani N. Therapeutic se-lective adsorption of anti-DNA antibody using dextran sulfatecellulose column (Selesorb) for the treatment of systemic lu-pus erythematosus. Ther Apheresis 1998;2:8–24.

3. Yoshida M, Tamura Y, Yamada Y, Yamawaki N, YamashitaY. Immusorba TR and Immusorba PH: Basics of design andfeatures of functions. Ther Apheresis 1998;2:185–92.

4. Fadul JEM, Danielson BG, Wikstrom B. Reduction of plasmafibrinogen, immunoglobulin G, and immunoglobulin M con-centrations by immunoadsorption therapy with tryptophanand phenylalanine adsorbents. Artif Organs 1996;20:986–90.

5. Kessler SW. Rapid isolation of antigens from cells with astaphylococcal protein A-antibody adsorbent: Parameters ofthe interaction of antibody-antigen complexes with protein A.J Immunol 1975;115:1617–24.

6. Muller-Derlich J, duMoulin A, Spaethe R. Specific removal ofhuman immunoglobulin with IgG-Therasorb. Abstract. ArtifOrgans 1993;17:546.

7. Schossler W, Hiepe F, Coupek J, Apostoloff E. Binding of C1qand DNA to support materials by means of a new couplingprocedure. Biomater Artif Cells Artif Organs 1990;18:657–63.

8. Euler HH, Schroeder JO, Harten P, Zeuner RA, GutschmidtHJ. Treatment-free remission in severe lupus erythematosusfollowing synchronization of plasmapheresis with subsequentpulse cyclophosphamide. Arthritis Rheum 1994;37:1784–94.

9. Sprenger KBG, Huber K, Kratz W, Henze E. Nomograms forthe prediction of patients’ plasma volume in plasma exchangetherapy from height, weight, and hematocrit. J Clin Apheresis1987;3:185–90.

10. Langone JJ. Protein A of staphylococcus aureus and relatedimmunoglobulin receptors produced by streptococci andpneumococci. Adv Immunol 1982;32:157–252.

11. Sasso EH, Silverman GJ, Mannik M. Human IgA and IgGF(ab8)2 that bind to staphylococcal protein A belong to theVHIII subgroup. J Immunol 1991;147:1877–83.

12. Bevan DJ, Carey BS, Lea CK, Vaughan RW, Rigden SPA,Higgins RM, Hendry BNM, Koffman CG, Bewick M. Anti-body removal and subsequent transplantation of a highly sen-sitised paediatric renal patient. Transplant Int 1996;9:155–60.

13. Braun N, Gutenberger S, Erley CM, Risler T. Immunoglob-ulin and circulating immune complex kinetics during immu-noadsorption onto protein A sepharose. Transfus Sci 1998;19(Suppl.):25–31.

14. Kadar JG, Parusel M, Spath PJ. Immunological effects oftherapeutic immunoadsorption with respect to biocompatibil-ity. Transfus Sci 1998;19(Suppl.):9–23.

15. Braun N. Expression of adhesion-molecules and activationmarkers on lymphocytes and monocytes during hemodialysis.Blood Purif 1997;15: 61–76.

16. Bonomini V, Vangelista A, Frasca GM, Nanni-Costa A,Borgnino LC. Effect of plasmapheresis on cellular immunityabnormalities in patients with systemic lupus erythematosus.Clin Nephrol 1984;22:121–6.

17. Vangelista A, Frasca GM, Nanni-Costa A, Bonomini V. Ef-fect of plasmapheresis on E-rosette receptors in SLE patients.J Clin Apheresis 1986;3:100–2.

18. Suzuki K, Taman J, Matsuki Y, Hidaka T, Ishizuka T,Kawakami M, Yabuki T, Kutsuki H, Nakamura H. Anti-dsDNA antibody kinetics during in vivo apheresis in systemiclupus erythematosus patients and in an in vitro apheresismodel. J Clin Apheresis 1996;11:211–6.

IMMUNOADSORPTION 245

Therapeutic Apheresis, Vol. 3, No. 3, 1999