Pyrogen retention by highly permeable synthetic membranes during in vitro dialysis

Pyrogen Retention by Highly Permeable SyntheticMembranes During In Vitro Dialysis

*Gerhard Lonnemann, ‡Luisa Sereni, †Horst-Dieter Lemke, and ‡Ciro Tetta

*Gemeinschaftspraxis, Eickenhof, Langenhagen; †Membrana Research, Obernburg, Germany; and ‡Clinical andLaboratory Research Department, Bellco, Mirandola, Italy

Abstract: Pyrogen permeability of the new highly perme-able synthetic membrane polyethersulfone (DIAPES) wascompared to polysulfone in vitro dialysis experiments withheparinized human donor blood in the blood compart-ment. After sterile dialysis for 5 min, dialysate was con-taminated with a culture filtrate of Pseudomonas aerugi-nosa using high and moderate challenge doses (Limulusassay reactivity 20,000 EU/ml and 50 EU/ml, respectively).Whole blood samples were separated from the blood com-partment during the sterile (5 min) and contaminated (60min) phases of dialysis and incubated for 6 h at 37°C.Blood samples were lysed, and interleukin-1� and tumornecrosis factor � were measured by specific ELISAs. Mod-erate dialysate contamination (50 EU/ml) did not inducedetectable cytokine production in whole blood. Highchallenge dose (20,000 EU/ml) induced whole blood inter-leukin-1� and tumor necrosis factor � production in theblood compartment, which was higher with DIAPES thanwith polysulfone after 30 min. After 60 min, membrane-

dependent differences were no longer detectable. Pyrogenconcentrations in the dialysate decreased with time indi-cating adsorption of cytokine-inducing substances to thedialyzer membrane. Pyrogens adsorbed to dialyzer mem-branes were resuspended during recirculation of sterilephosphate-buffered saline in the dialysate compartmentand retained cytokine-inducing activity as seen from wholeblood incubation experiments. DIAPES and polysulfoneadsorbed pyrogens in the presence of whole blood. Pyro-gen adsorption to the membrane polymer and/or to theprotein coat on the membrane prevented the passage ofpyrogens in the presence of moderately contaminateddialysate. High grade dialysate contamination causedbreakthrough of pyrogens into the blood with DIAPESand polysulfone. In order to reduce the risk of a dialysis-dependent inflammatory response, dialysate of high bac-teriological quality (ultrapure dialysate) should be man-datory. Key Words: Pyrogens—Hemodialysis—Cytokines—Inflammation.

Gram-negative bacteria, particularly those of thePseudomonas strain, the commonest contaminantsof dialysate, release a variety of substances includingthe classical endotoxin but also small molecular sub-stances such as peptidoglycans and muramyldipep-tides that are structurally and biologically unrelatedto endotoxin (1). Because of the biological activityshared by all these substances, they are collectivelynamed pyrogens or “cytokine-inducing substances”(CISs). CISs induce proinflammatory cytokines suchas interleukin-1� (IL-1�), or tumor necrosis factor �(TNF�) in human peripheral blood mononuclearcells (PBMCs) during in vitro cell culture. The in

vitro incubation of isolated mononuclear cells hasbeen introduced as a sensitive system to detect alldifferent CISs (2). In contrast, the limulus amoebo-cyte lysate (LAL) test only detects the intact endo-toxin and endotoxin fragments bearing the lipid Asubunit but not other relevant CISs such as peptido-glycans.

Recent in vitro studies show that the synthetichighly permeable membranes such as polyamide andpolysulfone are less permeable to CIS derived fromdifferent Pseudomonas strains as compared to low-flux as well as highly permeable cellulosic mem-branes (2,3). In these studies, CISs appearing in theblood compartment of the in vitro dialysis systemswere not inhibited by polymixin B (PMX-B), a lipo-polysaccharide (LPS)–specific antibiotic, suggestingthat low molecular weight pyrogens penetrating dia-lyzer membranes are predominantly different fromthe intact endotoxin.

Received March 2001; revised May 2001.Address correspondence and reprint requests to Dr. Ciro Tetta,

Clinical and Laboratory Research Department, Bellco S.p.A.,Via Camurana 1, 41037 Mirandola, Italy. E-mail: [email protected]

Artificial Organs25(12):951–960, Blackwell Science, Inc.© 2001 International Society for Artificial Organs

951

DIAPES is a new synthetic polyethersulfone usedfor manufacturing highly permeable membrane inhollow-fiber dialyzers. DIAPES is different fromknown synthetic highly permeable membranessuch as polyamide or polysulfone with respect toreduced membrane thickness (30 �m as comparedto 40–50 �m) that results in a higher hydraulic per-meability in water (4). Reduced thickness of thissynthetic highly permeable membrane may be asso-ciated with improved dialyzer clearance of mole-cules in the molecular weight range of 10,000 to20,000 daltons (e.g., �2-microglobulin, AGEs) in thepresence of a clinically acceptable loss of albumin(5,6).

The aim of the present study was to compare theCIS permeability of DIAPES to that of polysulfoneunder the conditions of in vitro hemodialysis withwhole human donor blood in the closed loop bloodcompartment and bicarbonate dialysate contami-nated with a culture filtrate of Pseudomonas aerugi-nosa. The closed loop dialysate compartment con-tained bicarbonate dialysate contaminated with aculture filtrate of Pseudomonas aeruginosa. Trans-membrane passage of CIS was determined bymeasuring the cytokine production in whole bloodsamples drawn from the blood compartment atvarious time points after contamination of dialy-sate. Two challenge doses of Pseudomonas aerugi-nosa culture filtrate were used in order to test theCIS permeability of the 2 membranes under theconditions of high and moderate dialysate contami-nation.

MATERIALS AND METHODS

In vitro dialysis circuitWe used a previously described model of in vitro

dialysis (3). Standard dialysis tubing sets were con-nected to a dialyzer under sterile conditions. Theblood compartment (BC, volume 150 ml) and thedialysate compartment (DC, volume 200 ml) wererinsed with 2 L of pyrogen-free saline each. Afterrinsing, the saline was replaced by pyrogen-free bi-carbonate dialysate in the DC and by human bloodin the BC. Heparinized fresh human donor bloodwas diluted 1:2 in ultrafiltered tissue culture medium(RPMI 1640, Sigma Chemical Co., St. Louis, MO,U.S.A.). The diluted blood (hematocrit 20–24%)was recirculated for 5 min in the BC in the presenceof sterile bicarbonate dialysate in the DC. Followingthis sterile control phase, dialysate was contaminatedby injecting the culture filtrate of Pseudomonas ae-

ruginosa into the DC (70–120 ml), replacing an equalvolume of bicarbonate dialysate. Different volumesof culture filtrate reflected adjustment to thestrength of the challenge material according to thecontent of LAL-reactive material (see below). Con-taminated in vitro dialysis was performed for 1 hwith a blood flow of 150 ml/min and a countercur-rent dialysate flow of 300 ml/min. The BC and DCwere tightly connected without air in the small res-ervoirs in order to prevent net volume shifts acrossthe dialyser membrane due to ultrafiltration or back-filtration. Two identical dialysis circuits were run inparallel for each experimental design (see below) inorder to compare the 2 synthetic hollow fiber mem-branes: �-sterilized DIAPES (DIAPES HF800,Membrana GmbH, Wuppertal, Germany; dialyzerBLS 814, 1.4 m2, Bellco, Mirandola, Italy) and eth-ylene oxide–sterilized polysulfone (Fresenius AG,Bad Homburg, Germany; dialyzer BLS 627, 1.3 m2,Bellco).

Challenge materialPseudomonas aeruginosa (ATCC 9027, Medical

and Veterinary Supplies Ltd., Botolph Claydon,Buckingham, U.K.) were grown in bicarbonate dial-ysate until log-phase of growth. The culture wassonicated (30 min at room temperature) and filteredconsecutively through filters (Minisart, Sartorius,Goettingen, Germany) with decreasing pore sizeswith the final filter having a cutoff of 0.2 �m. Thestrength of the Pseudomonas aeruginosa culture fil-trates was assessed in 2 ways: first, by using the LALassay (BioWhittaker, Walkersville, MD, U.S.A.) tomeasure the content of endotoxin and lipid A–con-taining endotoxin fragments and second, by incubat-ing the culture filtrates with whole human donorblood (diluted 1:2 in RPMI 1640, Sigma ChemicalCo.) for 6 h at 37°C in an atmosphere with 5% CO2

to test the cytokine-inducing activity of the chal-lenge material. In addition, Pseudomonas aeruginosaculture filtrates were also preincubated with eitherPMX-B (Sigma, in sterile phosphate buffered sa-line [PBS] at final concentration 36 �g/ml) or withPBS for 1 h at room temperature before addition tothe whole blood assay. Since PMX-B is an anti-biotic binding specifically to the lipid A subunit ofendotoxin, PMX-B–dependent inhibition of cyto-kine production in the whole blood assay indicatesthe presence of endotoxin or lipid A–bearingendotoxin fragments. Pseudomonas aeruginosa cul-ture filtrates were freshly prepared for every set ofexperiments, and the challenge dose used to con-taminate the dialysate in the in vitro experiments

G. LONNEMANN ET AL.952

Artif Organs, Vol. 25, No. 12, 2001

was adjusted according to the content of LAL-reactive material.

Experimental design

Design 1In vitro dialysis experiments were performed with

2 different doses of dialysate contamination as indi-cated by the amount of LAL-reactive material whichwas either 20,000 or 50 EU/ml. In order to test thekinetics of CIS transfer across the dialyzer mem-branes, samples from the BC and DC were takenafter 5 (sterile control), 30, and 60 min of recircula-tion. In additional experiments, the BC and DCsamples were taken more frequently (at 5, 10, 15,and 30 min). Whole blood samples taken from theBC (2 ml each) were divided into 2 aliquots of 1 mleach and incubated for 6 h at 37°C with equal vol-umes of either pyrogen-free PBS or with the dialy-sate sample taken at the same time from the DC.Using this experimental approach, it was possible tomeasure the appearance of CIS in the blood com-partment by comparing samples taken during thesterile control phase to those obtained during invitro dialysis with contaminated dialysate. In addi-tion, at every time of sampling the spontaneous aswell as the dialysate-induced cytokine production inwhole blood was assessed. Increased spontaneouscytokine production in whole blood during in vitrodialysis with contaminated dialysate indicated pas-sage of CIS across the dialyzer membrane. Cytokineproduction in BC samples during incubation withsamples of contaminated dialysate proved the viabil-ity of the cytokine-producing PBMC during closedloop in vitro dialysis and indicated the strength ofthe challenge material.

In order to describe further the type of solubleCIS appearing in the BC, whole blood samples ob-tained from the BC after 30 min of in vitro dialysiswith contaminated dialysate were centrifuged toseparate plasma. These plasma samples were thenpreincubated with or without PMX-B (as describedabove) and subsequently incubated (6 h at 37°C)with fresh human blood from the same donor thathad not been dialyzed. An inhibitory effect ofPMX-B would indicate the passage of endotoxin orlipid A–bearing endotoxin fragments across the dia-lyzer membranes.

Design 2Additional experiments were performed in order

to evaluate the possibility of readsorption of biologi-cally active CISs to the dialyzer membrane after pas-sage into the BC. In vitro dialysis was done as out-

lined above with the highest challenge dose in thedialysate (endotoxin content 20,000 EU/ml). After30 min of in vitro dialysis with contaminated dialy-sate, the BC and DC were rinsed with 2 L of PBS(pH 7.2) each. Fresh whole human donor blood (di-luted 1:2 in RPMI 1640) replaced PBS in the BC andrecirculated against sterile PBS in the DC for 1 h.Samples from the BC and DC were taken after 5, 15,30, and 60 min. Whole blood samples from the BCwere split into 2 aliquots of 1 ml each and incubatedfor 6 h with equal volumes of either sterile PBS orthe corresponding DC sample taken at the sametime.

Cytokine assaysAll whole blood samples were incubated for 6 h at

37°C in an atmosphere containing 5% CO2. Thus,spontaneous and dialysate-induced cytokine produc-tion was assessed. After incubation, samples werelysed in Triton X-100 (BDH England Chemical,Poole, U.K., at final concentration of 0.5%) andstored at −70°C until measurement of cytokines us-ing ELISAs. Immunoreactive TNF� and IL-1� weredetermined using commercially available ELISAs(Quantikine, R&D Systems, Minneapolis, MN,U.S.A.). The TNF� ELISA had a sensitivity of 4.4pg/ml, and the intraassay as well as the interassayvariation for known concentrations of recombinantTNF� was less than 6% (as given by R&D Systems).The lower detection limit of the IL-1� ELISA was1 pg/ml with an intraassay and interassay varia-tion for known amounts of recombinant IL-1� ofless than 4%. All samples were assayed in dupli-cates undiluted and in a 1:10 dilution. Measuredconcentrations in the range of the standard curveswere used to calculate the cytokine content in thesamples.

StatisticsMeasured cytokine concentrations are given in

nanograms per milliter of whole blood. Results areexpressed as means ± standard deviation. Paired Stu-dent’s t-test was applied using the SPSS softwarepackage. A p value of less than 0.05 was accepted assignificant.

RESULTS

Cytokine-inducing activity of Pseudomonasaeruginosa culture filtrate

The Pseudomonas aeruginosa filtrate inducedsignificant amounts of IL-1� and TNF� down to a

PYROGENS AND HIGHLY PERMEABLE MEMBRANES 953


dilution of 1:1,000 (Fig. 1). The 1:10,000 dilu-tion did not induce cytokine production above nega-tive controls. PMX-B reduced IL-1� and TNF�production by more than 50% of the untreatedsamples of the challenge material indicating thatthe majority of the cytokine-inducing activity pres-ent in the Pseudomonas aeruginosa filtrate wasdue to endotoxin or lipid A–bearing endotoxin frag-ments.

Kinetics of cytokine-inducing substance passageacross DIAPES and polysulfone membranes

Whole blood IL-1� production, after 5 min of ster-ile dialysis (control), was 0.081 ± 0.005 ng/ml withDIAPES and 0.089 ± 0.008 ng/ml with polysulfone.Thirty minutes after challenge of the dialysate(20,000 EU/ml), whole blood IL-1� production inthe BC increased significantly to 2.73 ± 0.25 ng/mlwith DIAPES (p � 0.0001 versus sterile control) butnot with polysulfone (1.02 ± 0.67 ng/ml, p � 0.074

versus sterile control). This difference was no longerapparent after 60 min of contaminated in vitro di-alysis because IL-1� production in the BC decreasedto 0.74 ± 0.53 ng/ml with DIAPES but tended toincrease with time to 1.32 ± 0.30 ng/ml with polysul-fone. Measuring TNF� instead of IL-1� in the wholeblood samples taken from the BC gave a similar re-sult.

When samples of the contaminated dialysate inthe DC were added to whole blood samples takenfrom the BC at the same time, an interesting ob-servation was made. Dialysate samples taken after60 min of contaminated in vitro dialysis induced 1.57± 0.17 ng/ml IL-1� in whole blood with DIAPESand 5.00 ± 2.79 ng/ml with polysulfone. The cyto-kine-inducing activity at 60 min tended to de-crease in respect to that observed at 30 min withoutreaching statistical significance due to the highstandard deviations with both tested dialyzers(DIAPES, dialysate at 30 min 7.38 ± 4.02 ng/ml IL-

FIG. 1. The graphs show the dose response of Pseudomonas aeruginosa filtrate-induced IL-1� (A,C) and TNF� (B,D) production duringwhole blood incubation for 6 h at 37°C. Samples were obtained from the DC of DIAPES (A,B) and polysulfone (C,D). Results of n = 3experiments (for each membrane) are given as means ± SD [samples from the DC at 30 min without PMX-B (�), at 60 min without PMX-B( ), at 30 min with PMX-B ( ), and at 60 min with PMX-B ( )].



1�; polysulfone, dialysate at 30 min 12.67 ± 7.23 ng/ml IL-1�).

Donor-dependent variability in whole bloodcytokine production

The experiments summarized in the precedingsection were carried out using whole blood from thesame healthy donor on different days. To investigatethe relevance of donor-dependent differences,additional experiments were performed using bloodfrom 3 different donors. Whole blood samples wereseparated from the BC after 5 (sterile control), 10,15, and 30 min of in vitro dialysis with highly con-taminated dialysate (LAL-reactive material 20,000EU/ml) (Fig. 2). A wide variability in the wholeblood response was seen among the three differentdonors (p > 0.05). Significant IL-1�- and TNF�-inducing activities were detectable in the BC withboth synthetic membranes already after 10 min ofcontaminated dialysis and increased further to reacha plateau after 15 min. No significant differencesbetween DIAPES and polysulfone were observed

at any of the time intervals due to the large interin-dividual variability in the measured cytokine pro-duction.

Effect of polymyxin B on cytokine-inducingactivity penetrating dialyzer membranes

In order to determine whether CIS activity ap-pearing in the blood compartment was related toendotoxin or lipid A–containing fragments thereof,plasma obtained from whole blood samples in theBC after 30 min of in vitro dialysis with highly con-taminated dialysate (LAL-reactive material in theDC 20,000 EU/ml) was preincubated with PMX-B.IL-1�–inducing activity appearing in the BC ofboth dialyzers was markedly (approximately 60%)inhibited by PMX-B indicating that lipid A–contain-ing endotoxin fragments derived from Pesudomonasaeruginosa were able to penetrate DIAPES as wellas polysulfone membranes (range with DIAPESand polysulfone 3230–4360 pg/ml, range withPMX-B 723–837 pg/ml). Interestingly, the PMX-Beffect was more pronounced on IL-1� than on TNF�

FIG. 2. Kinetics of the appearance of IL-1�– and TNF�–inducing activities in BC (full circles) of the in vitro dialysis circuit aftercontamination of the dialysate with Pseudomonas aeruginosa culture filtrates (LAL-activity: 20,000 EU/ml) using blood from three differentdonors. IL-1� and TNF� induced by dialysate samples are also depicted (full squares). Panels A to F: DIAPES; panels G to N:polysulfone.



production. There were no significant membrane-dependent differences in the PMX-B effect for bothcytokines.

Moderate grade contamination of dialysate doesnot induce whole blood cytokine production

In an additional set of in vitro dialysis experi-ments, the challenge dose was reduced to a concen-tration of LAL-reactive material of 50 EU/ml (Fig.3). Dialysate samples separated 30 min after con-tamination induced significant (p < 0.0001) pro-duction of IL-1� and TNF� in whole blood withDIAPES (for IL-1� 2.89 ± 0.77 ng/ml, for TNF� 0.35± 0.03 ng/ml) and polysulfone (for IL-1� 3.43 ± 0.57ng/ml, for TNF� 0.47 ± 0.06 ng/ml). However, wholeblood samples taken from the BC at the same timedid not produce significant amounts of both IL-1�(DIAPES 0.01 ± 0.002 ng/ml, polysulfone 0.006 ±0.007 ng/ml) and TNF� (DIAPES 0.009 ± 0.007 ng/ml, polysulfone 0.008 ± 0.008 ng/ml, p > 0.05 versuswhole blood incubated in sterile PBS only) (Fig. 3).These data indicate that a degree of dialysate con-tamination still high (endotoxin concentration 50EU/ml) compared to the clinical situation in mosthemodialysis units did not induce detectable cyto-kine production in the blood side when the synthetichighly permeable membranes DIAPES or polysul-fone were used.

Adsorption of cytokine-inducing substance todialyzer membranes

CIS added to the dialysate with the Pseudomonasaeruginosa culture filtrate could be removed fromthe fluid phase of the dialysate by adsorption ontothe dialyzer membrane and/or the tubings. The phe-

nomenon of adsorption could explain the decreaseof CIS in dialysate with time. Similarly, CIS appear-ing in the BC could also bind to the synthetic mem-brane either by direct interaction or via binding toplasma proteins that coated the blood side surface ofthe membrane. In order to evaluate any possiblereadsorption of protein-bound CIS to the dialyzermembrane, additional experiments were performed,and the results are depicted in Table 1. In vitro di-alysis experiments with highly contaminated dialy-sate and whole blood in the BC were performed for30 min. Then, the BC and DC were rinsed with ster-ile PBS, and fresh donor blood was added to the BCand recirculated for 60 min against sterile PBS in theDC. (See Design 2 under Materials and Methods.)Fresh blood samples produced small but significantamounts of IL-1� compared to controls obtained af-ter sterile recirculation (Time 0, 0.015 ± 0.007 ng/ml).The spontaneous IL-1� production in the BC withDIAPES was for Donor 1, 0.07 ± 0.07 ng/ml, p �0.234; for Donor 2, 0.24 ± 0.05 ng/ml, p � 0.0001; andfor Donor 3, 0.33 ± 0.15 ng/ml, p � 0.017; the pro-duction with polysulfone for Donor 1 was 0.03 ± 0.01ng/ml, p � 0.068; for Donor 2, 0.02 ± 0.01 ng/ml, p �0.889; and for Donor 3, 0.05 ± 0.01 ng/ml, p � 0.016.According to these data, CIS penetrating the intactsynthetic dialyzer membranes during contaminatedin vitro dialysis seem to be adsorbed to the proteinlayer coating the blood side surface of the mem-brane. Protein-bound CISs are only partially re-moved during the rinsing procedure with PBS andretain cytokine-inducing activity when fresh blood isrecirculated in the BC. When PBS samples obtainedfrom the DC after the rinsing procedure were incu-bated with whole blood samples from the BC, high

FIG. 3. No appearance of IL-1� and TNF�–inducing activities in BC (full circles) of the dialysis circuit when the dialysate was moderatelycontaminated with Pseudomonas aeruginosa culture filtrate (LAL activity: 50 EU/ml) using whole blood from two different donors; IL-1�and TNF�–inducing activities in DC are given in full squares. Panels A to D: DIAPES; panels E to H: polysulfone.



concentrations of IL-1� were produced (Table 1).These data indicate that CIS adsorbed onto themembrane could be re-uptaken and transferred byprotein-free solutions such as PBS.

DISCUSSION

Bacterial contamination of bicarbonate buffereddialysate is still a common problem in modern he-modialysis units (1). The clinical relevance of con-taminated dialysate-related triggering of monocyteactivation and cytokine induction in chronic inflam-mation has been reviewed recently (7). Transmem-brane passage of pyrogens derived from contami-nated dialysate across the intact dialyzer membranemay induce leukocyte activation, leading to cytokineproduction and consequently to increased levels ofacute phase proteins (e.g., C-reactive protein, CRP)indicating a state of inflammation in chronic hemo-dialysis patients. Even in the absence of a clinicallyobvious inflammatory response with fever and hypo-tension, the subclinical level of inflammation as in-dicated by increased mononuclear cell cytokine pro-duction and raised levels of CRP may contribute tochronic inflammatory diseases such as tissue damagein the course of �2-microglobulin amyloidosis, pro-tein catabolism, and arteriosclerosis which are asso-ciated with long-term hemodialysis therapy (1). Be-cause bacteria-derived pyrogens are the most potentinducers of proinflammatory cytokines, hemodialysismembranes, especially highly permeable mem-branes, should provide a safe barrier to prevent pas-sage of pyrogens from dialysate into the patient’sblood.

In the present study, we tested the pyrogen reten-

tion by DIAPES, a new synthetic polyethersulfone,which has a higher hydraulic permeability than otherknown highly permeable membranes (4), and com-pared it to that of highly permeable polysulfone. Thereason for choosing polysulfone for comparison wasthe fact that this highly permeable synthetic mem-brane was shown to have a high adsorbing capacityfor CISs derived from bacterial cultures (2,3) as wellas for purified endotoxin (8). Particular attentionwas devoted to the appropriateness of the bacterialsource and the constancy of the challenge dose aswell as to the issue of donor-dependent variability inthe cytokine response of whole blood to pyrogenchallenge. We tested the pyrogen permeability of the2 dialyzer membranes using 2 closed loop in vitrodialysis circuits in parallel experiments at the sametime and using the same donor blood and the samepreparation of challenge material to contaminate thedialysate. In order to closely mimic the in vivo situ-ation we used a culture filtrate of Pseudomonas ae-ruginosa, a common dialysis water contaminant,grown in bicarbonate dialysate to contaminate thedialysate. Culture filtrates of Pseudomonas aerugi-nosa were always prepared fresh on the day beforethe dialysis experiments were done. The pyrogenicstrength of the filtrate was determined in 2 ways:first, by using the LAL assay to measure endotoxinconcentrations and second, by incubating serial dilu-tions of the Pseudomonas aeruginosa culture filtratewith whole donor blood to measure its cytokine-inducing activity. The strength of the various culturefiltrates was highly reproducible and induced mea-surable cytokine production down to a dilution of1:1,000 (Fig. 1). This cytokine-inducing activity wasinhibited by preincubation with PMX-B by 50 to80% indicating that more than 50% of the cytokine-inducing activity was due to endotoxin or lipidA–containing endotoxin fragments. The challengedose used in the in vitro dialysis experiments wasadjusted according to the content of LAL-reactivematerial to either high or moderate grade contami-nation (LAL reactivity 20,000 and 50 EU/ml, respec-tively). It should be emphasized that standards forthe bacteriological quality of dialysate allow a maxi-mal endotoxin concentration of 0.5 EU/ml (1). Thus,the challenge dose of 50 EU/ml was relatively high.In addition, the Pseudomonas aeruginosa filtratecontains bacterial products that are not detected bythe LAL assay but which account for 20 to 50% ofthe cytokine-inducing activity.

We used heparinized whole human donor bloodcontaining all physiologic enhancers of pyrogens(such as LPS-binding protein) and inibitors such asbactericidal/permeability inhibitory protein (BPI)

TABLE 1. IL-1 �-inducing activity of pyrogens adsorbedto synthetic dialyzer membranes

Donor

Time (min)

0 5 15 30 60

DIAPES (ng/ml) 1 0.021 0.71 0.68 0.58 0.622 0.008 0.18 0.24 4.29 1.153 0.016 1.8 2.51 1.98 1.04

Polysulfone 1 0.021 1.49 1.39 1.33 1.29(ng/ml) 2 0.008 0.64 0.57 0.57 0.59

3 0.016 0.43 0.54 0.51 0.39

After 30 minutes of in vitro dialysis with whole blood in BCand highly contaminated dialysate (LAL-activity: 20,000 EU/ml)in DC, both compartments were rinsed with sterile PBS. There-after, fresh blood from the same donor (diluted 1:2 in RPMI 1640)was added to BC and recirculated against PBS for 1 hour. Thetable depicts IL-1� production in whole blood samples taken fromBC at the indicated time points and incubated (6 hours at 37°C)with the corresponding PBS samples from DC. IL-1� productionin control samples (e.g. taken during sterile dialysis) was:0.015 ± 0.007 ng/ml.



and other nonspecific plasma proteins (9). The invitro incubation of whole blood has been shownpreviously to be a sensitive system for detectingall biologically active CISs of bacterial origin de-rived from purposefully contaminated dialysate(10,11). In the present study, the sensitivity of thewhole blood assay to detect CISs was further en-hanced by measuring total (cell-associated plus se-creted) cytokines in whole blood samples after celllysis by Triton X.

Finally, every in vitro dialysis experiment startedwith a sterile control phase of 5 min to assure thatthe system was indeed sterile as indicated by theabsence of detectable cytokine production in wholeblood samples taken from the dialysis circuit. A sec-ond important aspect is that during the first 5 min ofwhole blood recirculation in the BC under sterileconditions, plasma proteins interacted with the sur-face of the synthetic polymer and built a secondlayer, a process called protein coating of the dialyzermembrane. Protein coating was shown to reduce thepyrogen permeability of polysulfone membranes(12).

The first important result of our study is that in thepresence of high grade contamination (LAL reactiv-ity 20,000 EU/ml) in the dialysate, CISs appeared inthe BCs of both synthetic highly permeable dialyz-ers. This confirms previously published data on thepyrogen permeability of polysulfone as well as poly-amide when tested under similar in vitro hemodialy-sis conditions using whole human donor blood (3).However, the kinetics of CIS appearance in the BCwas different depending on the membrane used.Compared to polysulfone, DIAPES allowed signifi-cantly more IL-1�- and TNF�-inducing substancesto penetrate into the BC after 30 min of contami-nated in vitro dialysis. Thirty minutes later (at 60min), this membrane-dependent difference was nolonger apparent because with DIAPES whole bloodcytokine production in the BC decreased whereaswith polysulfone dialyzers it started lower but con-tinued to increase with time. One possible explana-tion for the earlier appearance of CIS in the BC ofDIAPES is its reduced membrane thickness (30 �m)compared to polysulfone (40 �m) because thinnermembranes facilitate diffusive transport across thedialyzer membrane. In addition, the thinner supportstructure of the synthetic polymer may also reducepyrogen-adsorbing areas such as hydrophobic do-mains resulting in reduced total pyrogen adsorbingcapacity and in an earlier breakthrough of CIS. It ismore difficult to explain why cytokine-producingblood cells in the BC of DIAPES are less activated30 min later during continued in vitro dialysis with

highly contaminated dialysate. Since the same do-nors’ blood showed increased cytokine productionafter 60 min compared to 30 min when polysulfonewas tested, an impaired responsiveness of the cyto-kine producing cells after 60 min of closed loop re-circulation is unlikely. One may speculate that CIScould be inactivated by plasma proteins such as BPI,but it remains unclear why such a BPI effect shouldbe more pronounced with DIAPES than with poly-sulfone. Protein coating of the inner membrane sur-face of the hollow fibers seems to be crucial. It ispossible that differences in the composition of the 2synthetic membrane polymers do not cause only dif-ferences in hydraulic permeability but also in theirability to adsorb pyrogens and plasma proteins caus-ing the protein coating of the membrane. Because ofthe various mechanisms by which plasma proteinsmay interact with the membrane polymers (e.g.,charge, hydrophobic interactions), the protein coaton DIAPES could be qualitatively and quantita-tively different from that on polysulfone. We specu-late that the protein layer on DIAPES forms slowerbut is able to reduce the passage of CIS into theblood compartment as well as the plasma coat onpolysulfone.

There is a second important aspect concerningCIS adsorption by synthetic membranes. The ex-periments (Design 2 under Materials and Methods)depicted in Table 1 clearly demonstrate that CISsadsorbed to synthetic membranes, either by directinteraction or via plasma proteins in the proteincoat, retain their IL-1�- and TNF�-inducing activi-ties.

Taken together, the in vitro dialysis experi-ments with whole blood in the BC demonstrate thatDIAPES and polysulfone show a considerable capa-bility to adsorb pyrogens derived from contaminateddialysate. Pyrogen adsorption either by direct inter-actions between the membrane polymers and the py-rogens or by plasma proteins coating the syntheticmembrane seems to be saturable, and breakthroughof pyrogens into the BC may occur.

In additional experiments, we used PMX-B to in-hibit the cytokine-inducing activity provided byendotoxin and lipid A–containing endotoxin frag-ments present in Pseudomonas aeruginosa culturefiltrate. We demonstrate that CISs appearing in theBC of polysulfone and DIAPES were significantlyinhibited by PMX-B, indicating that endotoxin orfragments thereof could penetrate both membranes.These data are in conflict with previously publisheddata showing that Pseudomonas-derived CISs pen-etrating intact dialyzer membranes were not inhib-ited by PMX-B (2). However, in the latter study,



TNF� was evaluated. The present investigationshows that the PMX-B effect is less pronounced inTNF� than in IL-1� production. Furthermore,differences in the procedure to produce the bacterialculture filtrate may also explain this discrepancy.In the present study, the bacterial culture wassonicated for 30 min before cascade filtration,which was not done in the previously publishedstudy (2). Exposure to ultrasound may disrupt bac-terial cell walls more intensively, thus yieldingmore endotoxin fragments of small to middle mo-lecular sizes.

The interindividual variability in the cytokineproduction in response to a given stimulus has alsobeen addressed in the present article. As demon-strated in Fig. 2 and Table 1, there is a remarkabledonor-dependent variation in the amount of cyto-kines produced by whole blood samples obtainedduring the in vitro dialysis experiments. So called“high responders” can be distinguished from “lowresponders,” which may reflect polymorphism inthe genes coding for IL-1� and TNF� (13), whichhas been shown for various cytokines not only inhealthy subjects but also in uremic patients (14). Onthe other hand, there was a very reproducible re-sponse in whole blood samples taken from thesame donor on different days, reflecting an intrain-dividual consistency in the cytokine response toCISs.

In experiments with a dialysate contaminationcontaining an LAL reactivity of 50 EU/ml in the DCand whole blood from 3 different donors in 3 inde-pendent experiments in the BC, no passage of CISwas detectable. There were no differences betweenDIAPES and polysulfone. We conclude that the ca-pacity of both membranes to adsorb CISs is suffi-cient to prevent whole blood activation providedthat the challenge concentration is not higher than50 EU/ml. Compared to the clinical situation wherethe endotoxin content in routine dialysate is lessthan 5 EU/ml in at least 90% of tested samples (11),50 EU/ml represents a high challenge dose. We con-clude that under routine hemodialysis conditionswith a degree of dialysate contamination within therecommended standards (bacterial growth less than2,000 colony forming units per milliter, endotoxinconcentration less than 5 EU/ml), the synthetic dia-lyzer membranes DIAPES and polysulfone providea sufficiently safe barrier to prevent the passage ofCISs contaminating routine dialysate into patients’blood.

In conclusion, the retention of pyrogens is a com-plex phenomenon that involves the interaction withthe polymer itself as well as formation of a CIS-

adsorbing protein coat on the inner surface of thehollow fibers. The present study reemphasizes thatspecific synthetic polymers such as polysulfone andthe new DIAPES are effective barriers to preventthe passage of bacterial products derived from rou-tine dialysate at the commonly observed degree ofendotoxin contamination (<5 EU/ml). However,given the high interindividual variability in the pro-duction of proinflammatory cytokines, the capacityof plasma to extract biologically active membrane-bound bacterial products, and the in vitro nature ofour experiments, one cannot completely exclude thepossibility of a pyrogenic reaction due to contami-nated dialysate in vivo. Especially when syntheticmembranes are used in modern techniques such ashemodiafiltration with high convective driving forcesacross the membrane, dialysate fluid of high micro-biological quality (ultrapure) should be mandatoryto reduce the risk of dialysis-dependent inflamma-tion.

REFERENCES

1. Lonnemann G. The quality of dialysate: An integrated ap-proach. Kidney Int 2000;58(Suppl 76):S112–9.

2. Lonnemann G, Behme TC, Lenzner B, Floege J, Schulze M,Colton CK Koch KM, Shaldon S. Permeability of dialyzermembranes to TNF alpha-inducing substances derived fromwater bacteria. Kidney Int 1992;42:61–8.

3. Schindler R, Krautzig S, Lufft V, Lonnemann G, Mahiout A,Marra MN, Shaldon S, Koch KM. Induction of interleukin-1and interleukin-1 receptor antagonist during contaminated in-vitro dialysis with whole blood. Nephrol Dial Transplant 1996;11:101–8.

4. Jaber BL, Gonski JA, Cendoroglo M, Balakrishnan VS,Razeghi P, Dinarello CA, Pereira BJG. New polyether sul-fone dialyzers attenuate passage of cytokine-inducing sub-stances from Pseudomonas aeruginosa contaminated dialy-sate. Blood Purif 1998;16:210–9.

5. Gerdemann A, Lemke HD, Nothdurft A, Heidland A, MünchG, Bahner U, Schinzel R. Low-molecular but not high-molecular advanced glycation end products (AGEs) areremoved by high-flux dialysis. Clin Nephrol 2000;54:276–83.

6. Nensel U, Roeckel A, Hillenbrand T, Bartel J. Dialyzer per-meability for low-molecular weight proteins: Comparison be-tween polysulfone, polyamide and cuproammonium rayondialyzers. Blood Purif 1994;12:128–34.

7. Stenvinkel P. Malnutrition and chronic inflammation as riskfactors for cardiovascular disease in chronic renal failure.Blood Purif 2001;19:143–51.

8. Bommer J, Becker KP, Urbaschek R. Potential transfer ofendotoxin across high-flux polysulfone membranes. J Am SocNephrol 1996;7:883–8.

9. Schindler R, Marra MN, Mckelligon BM, Lonnemann G, Sch-ulzeck P, Schulze M, Oppermann M, Shaldon S. Plasma levelsof bactericidal/permeability-increasing protein (BPI) and li-popolysaccharide-binding protein (LBP) during hemodialysis.Clin Nephrol 1993;40:346–51.

10. Nerad JL, Griffiths JK, Van Der Meer JW, Endres S,Poutsiaka DD, Keusch GT, Bennish M, Salam MA, Dinarello



CA, Cannon JG. Interleukin-1 beta (IL-1 beta), IL-1 receptorantagonist, and TNF alpha production in whole blood. J Leu-koc Biol 1992;52:687–92.

11. Lonnemann G. Assessment of the quality of dialysate.Nephrol Dial Transplant 1998;13(Suppl 5):17–20.

12. Lonnemann G, Schindler R, Lufft V, Mahiout A, Shaldon S,Koch KM. The role of plasma coating on the permeation ofcytokine-inducing substances through dialyser membranes.Nephrol Dial Transplant 1995;10:207–11.

13. Molvig J, Pociot F, Baek L, Worsaae H, Dall WogensenL, Christensen P, Staub-Nielsen L, Mandrup-PoulsenT, Manogue K, Nerup J. Monocyte function in IDDM pa-tients and healthy individuals. Scand J Immunol 1990;32:297–311.

14. Girndt M, Sester U, Deman E, Sester M, Kaul H, Koehler H.Interleukin-10 promoter polymorphism determines immunefunction in dialysis patients (Abstract). J Am Soc Nephrol2000;11:268A.



Pyrogen retention by highly permeable synthetic membranes during in vitro dialysis

Documents