Online Hemodiafiltration Versus Acetate-Free Biofiltration:A Prospective Crossover Study

*Feng Ding, *Peter Ahrenholz, †Roland E. Winkler, †Wolfgang Ramlow, †Michael Tiess,†Ann Michelsen, and †Wolfgang Patow

*BioArtProducts GmbH and †Dialyse-Gemeinschaft Nord e.V., Rostock, Germany

Abstract: Online hemodiafiltration (online HDF) and ac-etate-free biofiltration (AFB) are 2 innovative renal re-placement therapies. Convincing evidence has shown thatboth techniques are superior to conventional hemodialysisin many aspects. The aim of the present investigation wasto compare online HDF and AFB in 12 stable mainte-nance hemodialysis patients in a prospective, randomizedcrossover trial. Twelve stable dialysis patients, age 49.7 ±11.3 years and on dialysis for 83.5 ± 76.7 months, weretreated prospectively and randomly by either AFB, predi-lution HDF (pre-HDF), or postdilution HDF (post-HDF)for a total of 36 weeks using exclusively F60S high-fluxdialyzers. Routine blood biochemical tests, bone metabo-lism parameters, and clearance for both small and largermolecular weight substances were measured at defined in-tervals. During the trial period inter- and intradialysissymptoms, e.g., hypotensive episodes and intradialysis ar-terial blood gas analyses, were recorded. Both online HDFand AFB were well accepted by the overwhelming major-ity of patients and also by the dialysis staff. Pretreatmentsodium, total and ionized calcium, chloride, bicarbonate,and urea did not differ within or between the 3 treatmentgroups. Potassium increased slightly in HDF patientswhile phosphate and b2-microglobulin (b2-M) decreased inall groups. After dialysis, AFB patients exhibited a signifi-cantly higher bicarbonate concentration and lower potas-sium level when identical potassium concentrations in di-alysate were used. Patients receiving AFB manifested lessintradialysis partial pressure of oxygen drop and partialpressure of carbon dioxide rise than those on HDF treat-

ments. HDF treatments could afford higher single-pooland double-pool Kt/V, higher effective urea and b2Mclearance, and lower total interdialysis symptom scoresthan the AFB treatment method. While bone metabolismparameters did not differ between the 3 dialysis modali-ties, some parameters such as deoxypyridinoline in HDFand osteocalcin, pyridinoline, and deoxypyridinoline inAFB deteriorated at the end of the crossover study. Alu-minum concentration decreased progressively to aboutone-third of prestudy values at the end of the study with all3 treatments. AFB was associated with a lower predialysismean arterial pressure (MAP), a smaller drop in MAPduring treatment, and similar hypotension episodes com-pared with the 2 HDF treatments. Albumin concentrationshowed a trend to decrease during the first 2 months of thetrial period followed by a slight increase thereafter but stillsignificantly lower than initial value at the end of cross-over. Both online HDF and AFB share most of the fea-tures of optimal renal replacement therapy. Online HDF issuperior to AFB in such aspects as increased delivereddialysis dose both for small and larger molecular weighttoxins and less interdialysis symptoms. On the other hand,AFB is associated with a smaller effect on arterial bloodgas values and improved intradialysis hemodynamic toler-ance. Some dialysis-related symptoms and complicationsin the case of our AFB practice could be attributable,at least in part, to low dialysate calcium level. KeyWords: Online hemodiafiltration—Acetate-free biofiltra-tion—End-stage renal disease—Renal replacementtherapy.

Substantial progress in medical science and tech-nology has been achieved in the last 30 years. How-ever, patients on hemodialysis treatment still mani-fest a higher incidence of dialysis-related morbidity

and mortality. Indeed, acute complications, such ashypotension, muscle cramps, allergic reactions, andarrhythmia, as well as chronic complications, b2-microglobulin (b2M) such as associated amyloidosis,accelerated atherosclerosis, and malnutrition areprevalent in this special population. They affect theirquality of life considerably. Furthermore, as the ageand comorbidity of dialysis patients increase, thecomplexity of their disease management increases aswell. Any effort to improve the overall quality of

Received May 2001.Presented in part on the 65th birthday of Prof. Dr. Horst

Klinkmann, on May 7, 2000, in Rostock, Germany.Address correspondence and reprint requests to Dr. Peter

Ahrenholz, BioArtProducts GmbH, Rigaer Strasse 21, 18107Rostock, Germany. E-mail: az@dialysis-north.de

Artificial Organs26(2):169–180, Blackwell Publishing, Inc.© 2002 International Society for Artificial Organs

169

renal replacement therapies should be madewhereas innovative treatment options are welcome.

To date, 2 highly attractive treatment modalitiesbased on convective therapies are provided to thedialysis community: online hemodiafiltration (onlineHDF) and acetate-free biofiltration (AFB). The ad-vantages of these 2 techniques have been discussed(1,2). They exhibit a series of advantages such as alarger convective clearance for low and higher mo-lecular weight uremic toxins, an improved cardiovas-cular stability, better acid-base correction, and anoverall higher biocompatibility in terms of bloodcompatible endotoxin-adsorbing dialysis mem-branes, ultrapure dialysis, and substitution fluids.

It is well known that online HDF and AFB aresuperior to conventional bicarbonate hemodialysistechniques. Convincing data, however, still are lack-ing when comparing the 2 available convective di-alysis modalities. AFB, albeit also based on HDFtechnology, differs from online HDF in several as-pects such as technological concept, base of dialysateand substitution fluid, the magnitude of convectiveclearance, the degree to which metabolic acidosis isable to be corrected, and intradialysis electrolyte ki-netics.

The aim of the present investigation was to com-pare online HDF and AFB in 12 stable patients onmaintenance hemodialysis in a prospective, random-ized crossover study. Several clinical and biochemis-try parameters such as intra- and interdialysis symp-toms, hemodynamic tolerance, bone metabolicindices, dialysis efficacy, and arterial gas variationsduring hemodialysis sessions also were assessed.

SUBJECTS AND METHODS

PatientsTwelve stable dialysis patients, age 49.7 ± 11.3

years and on dialysis for 83.5 ± 76.7 months, wereincluded in this prospective crossover study. The pri-mary renal diseases of the 8 males and 4 femaleswere chronic glomerulonephrits (5 patients), dia-betic nephropathy (1 patient ), inherent nephropathy(2 patients), vasoculitis (1 patient), and of unknownorigin (3 patients). After providing informed con-sent, all participants were randomized and dividedinto 3 groups, each receiving 1 type of the followingtreatment modes: AFB, predilution HDF (pre-HDF), and postdilution HDF (post-HDF). The totalstudy duration was 36 weeks with each treatmentmode evaluated for 12 weeks. Each patient served ashis or her own control. Prior to the study, 5 patientswere treated by post-HDF with high-flux filters, 6 byhigh-flux hemodialysis, and 1 by low-flux hemodialy-sis.

Dialysis protocolAll online HDF treatments were carried out with

F60S high-flux filters (polysulfone membrane, sur-face area 1.3 m2, Fresenius Medical Care, Bad Hom-burg, Germany, single-use application). Tap waterwas treated with water softeners, charcoal filtration,and reverse osmosis. Blood flow rate was fixed at 250ml/min, and treatment duration (282.5 ± 29.9 min,ranging from 210 to 300 min) was unchanged duringthe whole trial period. HDF treatment was carriedout with Fresenius 4008 dialysis monitors.

The monitors mixed treated water volumetricallywith bicarbonate concentrate, yielding the followingfinal composition in mmol/L: Na+ 138, Cl− 110.5,Mg2+ 0.5, HCO3

− 32, acetate 2.0, K+ 3.0 or 4.0, andCa2+ 1.25 or 1.5. Dialysis fluid was filtered standardlythrough a polysulfone ultrafilter (Diasafe, FreseniusMedical Care), which was replaced every 8 weeks.The substitution fluid was filtered further by a sec-ond polysulfone filter (HDF-filter, Fresenius Medi-cal Care), which was used standardly for 50 sessionsor 6 weeks. Dialysate and infusion flow rate were setat 620 and 180 ml/min in pre-HDF and at 720 to 740and 60 to 80 ml/min in post-HDF, respectively.

AFB treatment was carried out with a Hospal In-tegra monitor (Hospal Medizintechnik, Munich,Germany). The dialysis bath during AFB treatmentcontained the following electrolyte composition inmmol/L: Na+ 139, Cl− 147, Ca2+ 1.5 or 1.75, Mg2+ 0.5,and K+ 3.0 or 4.0. Dialysate in AFB was completelybuffer-free, and acidosis was corrected with a 166mEq/L sodium bicarbonate solution as the substitu-tion fluid. Dialysate and infusion flow rate were setat 500 and 25 to 30 ml/min. The infusion flow ratewas programmed to ensure a postdialysis plasma bi-carbonate level between 27 and 30 mmol/L.

Clinical and biochemical testsAnalysis of pre- and postdialysis samples during

midweek treatment sessions were performed at 4week intervals for urea, creatinine, sodium, potas-sium, chloride, phosphate, and total and ionized cal-cium as well as b2M. In predialysis samples only he-matocrit, aluminum, albumin, bone-specific alkalinephosphatase (BAP), osteocalcin, pyridinoline(PYD), deoxypyridinoline (DPYD), and intact para-thormone (iPTH) were assessed.

Fistula blood gas levels were measured immedi-ately before dialysis, at every 20 min in the first hourof the dialysis session, and every hour thereafter un-til the end of the session. Blood gas was measuredusing an AVL Electrolyte and Blood-Gas Analyzer(AVL OMNI 4, AVL Medical Instrument AG, BadHomburg, Germany). In addition, bicarbonate in di-

F. DING ET AL.170

Artif Organs, Vol. 26, No. 2, 2002

alysate also was measured by titration (DL21 Titra-tor, Mettler Instrument AG, Switzerland). Albuminwas measured by kinetics nephelometry (BeckmanImage System, U.S.A.). Aluminum was determinedby atomic absorption spectroscopy (SpectrAA 300Z,Varian Co., Australia). The b2M was measured bymicroparticle enzyme immunoassay (AXSYM, Ab-bott Co. U.S.A.). Osteocalcin and iPTH were deter-mined by immunoenzymatic assay (Diagnostic Prod-ucts Co., CA, U.S.A.). The analytical sensitivity forosteocalcin was 0.1 ng/ml and for iPTH 0.7 pg/ml.BAP was assessed by agarose gel electrophoresis(Rep-Processor, Helena Co., U.S.A.). PYD andDPYD were measured using a modified isocratichigh-performance liquid chromatography (HPLC)method described previously by James et al. (3).

For postdialysis blood sampling, a slow flow/stoppump method recommended by the Dialysis Out-comes Quality Initiative (DOQI) was applied, i.e.,blood was drawn directly from the arterial needleafter slowing down the blood pump to 50 ml/min for20 s in order to eliminate a fistula recirculation ef-fect. Single-pool (Kt/Vsp) and double-pool (Kt/Vdp)Kt/V were calculated monthly using the second gen-eration of the Daugirdas equation.

Single pool:Kt/Vsp = −Ln~R − 0.008 × t!

+ ~4 − 3.5 × R! × UF/W (1)

Ln is the natural logarithm, R is the urea reductionratio, t the dialysis session length, UF the ultrafiltra-tion volume, and W corresponds to the patient’spostdialysis weight.

Double pool:Kt/Vdp = Kt/Vsp − ~0.6 × Kt/Vsp/t!

+ 0.03 (2)

Kt/Vsp is the single-pool Kt/V, and t is the dialysissession length.

Spent dialysate was collected during the wholetreatment session partially with a constant flow via abranch tube as previously described (4). The analysisof the solute concentration in total volumes of thecollected dialysate enabled us to calculate the sub-stance amount that really has been removed. As aconsequence, the effective clearance for the testedsubstance could be estimated appropriately by thefollowing equation:

Keff = Vd × Cd × Ln~C0/Ct!/$t × ~C0 − Ct!% (3)

Vd is the total dialysate volume including the totalultrafiltration volume UF, Cd is the urea concentra-tion in collected dialysate, Ln is the natural loga-rithm, C0 and Ct are the predialysis and postdialysis

blood urea concentrations, and t is the dialysis ses-sion length. We assumed the generation rate of ureato be negligible and Keff to be constant during thetreatment interval.

Medicinal drugs administered orally and intrave-nously, including erythropoietin (EPO), calcium,and antihypertensives, were recorded every 4 weeks.

Clinical symptomatologyAt the end of every dialysis session, patients were

asked to fill in a questionnaire. The questionnaireincluded questions about interdialysis symptoms andintradialysis symptoms, each consisting of the follow-ing 8 items: headache, cramps, itching, restless-legsyndrome, anorexia, thirst, dyspnea, angina, andtiredness. The severity of symptoms was subjectivelyexpressed by the patients on a scale from 0 to 3(absence of symptoms was 0 and the highest levelwas 3). Hypotensive episodes, defined as a bloodsystolic pressure below 90 mm Hg accompanied bygeneral discomfort and requiring therapeutic inter-vention, were also registered.

StatisticsData were expressed as mean ± SD. Calculations

were performed using the SPSS package (Version8.0, SPSS inc., IL, U.S.A.). Significant differencescorresponded to a p value of less than 0.05. Two-wayanalysis of variance and sequential Students-Newman-Keuls tests were used for normal data if adifference existed. For data not normally distributednonparametric, Kruskal and Wallis H tests wereadopted.

RESULTS

Both online HDF treatment modalities and AFBwere safe and well accepted by the overwhelmingmajority of patients and all dialysis personnel in ourunit. During the 36 weeks of the observed period andin more than 1,000 treatments, no clinically relevantpyrogenic reactions were observed. Three patientsdid not comply with the designed protocol. One pa-tient’s fistula failed during the second month of pre-HDF (third shift of the whole study). A second pa-tient refused to finish post-HDF and insisted thatAFB be tried because of his unbearable shoulderpain; however, his pain did not lessen with AFBtreatment. A third patient had a severe headacheaccompanied by poorly controlled hypertension atthe end of pre-HDF shift and dropped out of thestudy before starting post-HDF modality. As a re-sult, 12 patients underwent AFB and pre-HDF treat-ments, and 11 underwent the post-HDF mode as acomparison.

HEMODIAFILTRATION VERSUS BIOFILTRATION 171

Artif Organs, Vol. 26, No. 2, 2002

The set and actual dialysate composition are re-ported in Table 1. AFB patients were dialyzed withhigher ionized calcium, potassium, and lower sodiumin dialysate when compared with HDF patients.

The changes of pretreatment biochemical param-eters, during the period under study, are shown inTable 2. No marked changes were found in serumsodium, total and ionized calcium, chloride, bicar-bonate, and urea within each group (except for so-dium at the middle of pre-HDF) or between groups(except for bicarbonate between pre-HDF and AFBat middle of crossover). A slight increase of pretreat-ment potassium levels with time was detected in the2 HDF modalities, reaching statistical significanceonly in the post-HDF group at the end of the cross-over period. Compared to the HDF modes, AFBpatients’ potassium concentration turned out to beslightly lower, this difference reaching significance,however, only at the middle of the crossover periodbetween AFB and pre-HDF. Compared with theprestudy values, all 3 groups showed both a loweredpretreatment phosphate and reduced b2M serumconcentrations, the difference for phosphate reach-ing significance at the beginning and middle of thecrossover period in post-HDF, at the middle of thecrossover period in pre-HDF, and for b2M only inpre-HDF at the middle of the crossover period.

Once again, we could not detect any differenceamong the 3 treatment modalities.

Table 3 lists data on plasma biochemical param-eters and their changes before and after dialysis.Postdialysis potassium, ionized calcium, and bicar-bonate concentrations were significantly differentwith predialysis values in all 3 methods. When iden-tical potassium concentrations in dialysate wereused, AFB patients showed significantly lower post-dialysis serum potassium levels than those at HDFmodes. While pretreatment bicarbonate levels arecomparable, the AFB group manifested significantlyhigher postdialysis bicarbonate concentrations. Nodifference could be detected with respect to postdi-

alysis sodium and total and ionized calcium and chlo-ride concentrations among the 3 dialysis modalities.

The results of arterial blood gas during 1 typicaldialysis session are shown in Table 4 (percentage ofpredialysis values were used for partial pressures ofoxygen [PO2] and carbon dioxide [PCO2]). It is note-worthy that patients on AFB treatment presentedboth a drop of PO2 and a rise of PCO2 to a smallerdegree than those on HDF treatments. In fact, astatistically significant reduction of PO2 was docu-mented at 40 min, 120 min, and end in post-HDF andat 20 and 40 min in pre-HDF and was entirely absentin AFB. Similarly, a significant increase of PCO2

could be seen at 20, 40, and 60 min in post-HDF; at20, 40, and 60 min in pre-HDF; and only at 40 min inAFB.

The reduction rates of urea, creatinine, phosphate,and b2M with both corrected and uncorrected post-dialysis values were similar with the 3 dialysis mo-dalities (Table 5). By means of partial dialysate col-lection, AFB showed a lower effective clearance forurea and b2M than both HDF modalities numeri-cally but only significantly for b2M (Fig. 1). Com-pared to the other 2 modalities, post-HDF alwaysmanifested the highest numerical effective clearancefor all 4 substances under investigation, the molecu-lar weight of which ranged from 60 to 11,800. Inagreement with these clearance results, AFB treat-ments always provided lower Kt/Vsp and Kt/Vdp val-ues than HDF (Fig. 2) but significantly higher valuesthan other patients in our center not joining thestudy (Kt/Vsp 4 1.06 ± 0.30, Kt/Vdp 4 0.92 ± 0.27,p < 0.001 compared to each studied group, includingall patients with a diuresis of about 500–2,000 ml/day).

As shown in Table 6, most bone metabolism pa-rameters did not differ between post-HDF, pre-HDF, and AFB, except for DPYD at the fourthweek between post-HDF and AFB and at the eighthweek between pre-HDF and AFB. Plasma BAP, os-teocalcin, PYD, and iPTH values at the end of the

TABLE 1. Expected and actual dialysate composition in Post-HDF, Pre-HDF, and AFB

Post-HDF Pre-HDF AFB

Expected Actual Expected Actual Expected Actual

Sodium (mmol/L) 138.0 141.2 ± 2.8 138.0 141.4 ± 1.7 139.0 133.8 ± 5.6Potassium (mmol/L) 3.0 3.2 ± 0.5 3.0 3.1 ± 0.1 3.0 2.9 ± 0.2

4.0 4.1 ± 0.2 4.0 4.1 ± 0.1 4.0 4.1 ± 0.2Ionized calcium (mmol/L) 1.25 1.28 ± 0.03 1.25 1.28 ± 0.04 1.50 1.52 ± 0.08

1.50 1.56 ± 0.11 1.50 1.51 ± 0.03 1.75 1.73 ± 0.05Bicarbonate (mmol/L) 32–34 33.0 ± 0.5 32–34 32.2 ± 0.8 0 1.8 ± 0.5Bicarbonate (mmol/L) (titration) 32–34 35.4 ± 0.9 32–34 34.8 ± 0.9 0 0

HDF: hemodiafiltration, AFB: acetate-free biofiltration.

F. DING ET AL.172

Artif Organs, Vol. 26, No. 2, 2002

study for post-HDF and pre-HDF were similar com-pared to the corresponding prestudy values with theexception of DPYD which increased markedly inboth HDF groups.

By contrast, osteocalcin, PYD, and DPYD levelsincreased significantly in patients undergoing 12week AFB treatment while only iPTH and BAPwere still stable during this period. Aluminum con-centration decreased progressively to about one-third of the prestudy values at the end of the studywith all 3 treatments.

During the study period, the inter- and intradialy-sis symptom score remained stable in all 3 groups.While no difference existed among the 3 treatmentmodes concerning intradialysis symptom score, AFBpatients showed a higher interdialysis symptomscore than the 2 HDF patients (2.57 ± 2.93 in AFB,1.99 ± 2.49 in post-HDF, 1.88 ± 2.40 in pre-HDF,p 4 0.037 when AFB versus post-HDF, p > 0.05when AFB versus pre-HDF, and post-HDF versuspre-HDF). In contrast, when every individual symp-tom included in the questionnaire was analyzed, wecould not detect any statistical difference among the3 modalities, either for interdialysis or for intradialy-sis symptoms.

Throughout the study period, predialysis mean ar-terial blood pressures (MAP) in AFB, post-HDF,

and pre-HDF were kept stable. AFB was associatedwith a lower predialysis MAP (89.2 ± 17.7 mm Hg forAFB, 94.0 ± 16.5 mm Hg for post-HDF, 93.8 ± 14.7mm Hg for pre-HDF, p < 0.05 AFB versus the other2 groups), a smaller drop in MAP during treatment(Fig. 3), and more numerical but not significant hy-potension episodes (AFB 8.6%, post-HDF 6.1%,pre-HDF 5.8%, p > 0.05 when compared to eachother).

Interestingly, pretreatment albumin concentra-tions showed a trend to decrease during the first 2months of the trial period, followed by a slight in-crease thereafter, but remained still significantlylower than initial values at the end of the crossoverperiod (Fig. 4). When all participants were dividedinto an HDF or non-HDF group according to thetreatment they received before the study, those ex-posed to non-HDF previously exhibited a higher al-bumin level at the beginning (41.1 ± 4.4 g/L forthose on HD, 39.4 ± 3.4 g/L on HDF) and lowervalues at the end of the study (36.2 ± 2.1 g/L on HD,37.8 ± 1.8 g/L on HDF) compared with those ex-posed to HDF modes, indicating that albumindropped to a lesser extent in the latter group. Whenall participants were regarded as 1 group, albuminlevels still did not reach prestudy values at the end ofthe study (Fig. 5).

TABLE 2. Pretreatment biochemical parameter changes in AFB, pre-HDF, and post-HDF

Baseline Start of crossover Middle of crossover End of crossover

Sodium (mmol/L) Post-HDF 142.6 ± 1.9 142.8 ± 2.4 140.6 ± 2.2 141.1 ± 2.5Pre-HDF 141.8 ± 2.9 140.6 ± 2.1a 141.5 ± 2.3AFB 141.5 ± 2.5 141.2 ± 2.8 140.8 ± 2.2

Potassium (mmol/L) Post-HDF 5.4 ± 0.4 5.6 ± 0.4 5.5 ± 0.5 5.8 ± 0.4a

Pre-HDF 5.7 ± 0.5 5.7 ± 0.5b 5.7 ± 0.5AFB 5.4 ± 0.5 5.2 ± 0.3 5.5 ± 0.4

Chloride (mmol/L) Post-HDF 100.9 ± 5.5 97.3 ± 5.8 98.1 ± 2.8 97.7 ± 5.2Pre-HDF 97.9 ± 3.5 98.6 ± 5.6 97.8 ± 3.5AFB 96.5 ± 7.4 98.8 ± 5.8 96.8 ± 8.2

Ionized calcium (mmol/L) Post-HDF 1.19 ± 0.08 1.20 ± 0.10 1.17 ± 0.10 1.19 ± 0.11Pre-HDF 1.19 ± 0.09 1.18 ± 0.11 1.15 ± 0.12AFB 1.15 ± 0.11 1.15 ± 0.11 1.19 ± 0.13

Total calcium (mmol/L) Post-HDF 2.27 ± 0.23 2.33 ± 0.20 2.31 ± 0.21 2.33 ± 0.20Pre-HDF 2.31 ± 0.26 2.28 ± 0.27 2.36 ± 0.22AFB 2.35 ± 0.20 2.38 ± 0.23 2.38 ± 0.24

Bicarbonate (mmol/L) Post-HDF 22.5 ± 1.0 22.0 ± 1.5 22.6 ± 1.5 22.7 ± 1.9Pre-HDF 22.0 ± 2.2 21.9 ± 2.2b 23.4 ± 1.8AFB 22.8 ± 1.6 24.0 ± 2.3 23.0 ± 2.5

Phosphate (mmol/L) Post-HDF 2.35 ± 0.65 1.89 ± 0.46a 1.85 ± 0.32a 2.08 ± 0.56Pre-HDF 2.21 ± 0.59 1.86 ± 0.38a 2.01 ± 0.52AFB 2.34 ± 0.71 2.09 ± 0.55 2.04 ± 0.51

b2-microglobulin (mg/L) Post-HDF 30.1 ± 10.1 24.5 ± 7.3 26.4 ± 6.4 26.3 ± 7.9Pre-HDF 28.0 ± 10.7 24.4 ± 5.3a 25.2 ± 6.2AFB 27.4 ± 5.5 25.6 ± 6.4 25.9 ± 6.3

Urea (mmol/L) Post-HDF 20.7 ± 5.3 18.4 ± 4.5 20.1 ± 4.4 19.0 ± 4.7Pre-HDF 18.7 ± 4.3 19.7 ± 4.6 19.2 ± 4.3AFB 21.8 ± 5.0 22.2 ± 3.6 22.6 ± 3.5

a p < 0.05 compared to baseline value.b p < 0.05 compared to AFB value.HDF: hemodiafiltration, AFB: acetate-free biofiltration.

HEMODIAFILTRATION VERSUS BIOFILTRATION 173

Artif Organs, Vol. 26, No. 2, 2002

Throughout the study period for each dialysismethod, neither hematocrit level, dosage of EPO,nor important medication doses such as vitamin Dcompounds and oral antihypertensive drugs showedan apparent difference. In the AFB group, 6 patientshad an increased dry weight, and 3 showed a reducedweight. The patient numbers with increased and re-duced weight were 3 and 3 in the post-HDF groupand 3 and 1 in the pre-HDF group, respectively.Mean dry weight before and after the study, how-ever, did not differ within the 3 treatment methods.

DISCUSSION

Among all the organ replacement therapies, renalreplacement therapy is the most successful and com-monly used modality. This holds true for about 1million uremic patients worldwide presently sup-ported by this technique. This figure is still consid-ered to increase at a ratio of 9% annually. However,the success of dialysis as a long-term therapy forchronic patients does not exclude the fact that alarge proportion of dialysis patients still suffers from

TABLE 4. Arterial blood gas changes during AFB, pre-HDF, and post-HDF

Treatment modes Beginning 20 min 40 min 60 min 120 min End

PO2 Post-HDF 100 99.9 ± 0.1 93.6 ± 0.1a 95.5 ± 0.1 93.2 ± 0.1a 92.3 ± 0.1b

Pre-HDF 100 90.3 ± 0.1a 87.1 ± 0.1b 90.9 ± 0.2 91.3 ± 0.1 92.0 ± 0.1AFB 100 98.3 ± 0.1 97.0 ± 0.1 96.2 ± 0.1 99.4 ± 0.1 95.2 ± 0.1

PCO2 Post-HDF 100 105.8 ± 0.1a 108.2 ± 0.1b 108.5 ± 0.0c 113.0 ± 0.1c 106.5 ± 0.1Pre-HDF 100 106.7 ± 0.1a 108.9 ± 0.1a 106.3 ± 0.1a 106.1 ± 0.1 100.8 ± 0.1AFB 100 100.8 ± 0.1 105.7 ± 0.1a 102.6 ± 0.1 101.8 ± 0.1 100.2 ± 0.1

pH Post-HDF 7.40 ± 0.02 7.41 ± 0.2 7.41 ± 0.02 7.43 ± 0.02 7.44 ± 0.03 7.47 ± 0.02d

Pre-HDF 7.39 ± 0.03 7.40 ± 0.03 7.42 ± 0.03 7.43 ± 0.04 7.44 ± 0.04 7.48 ± 0.05d

AFB 7.41 ± 0.03 7.42 ± 0.04 7.42 ± 0.04 7.43 ± 0.04 7.46 ± 0.03 7.49 ± 0.05

a p < 0.05.b p < 0.01.c p < 0.001, compared to beginning value.d p < 0.05, compared to AFB value.AFB: acetate-free biofiltration, HDF: hemodiafiltration, PO2: partial pressure of oxygen, PCO2: partial pressure

of carbon dioxide.

TABLE 3. Pre- and postdialysis plasma biochemical parameter changes duringAFB, pre-HDF, and post-HDF

Before dialysis After dialysis

Sodium (mmol/L) Post-HDF 142.8 ± 2.4 143.2 ± 2.6Pre-HDF 141.3 ± 2.1 142.7 ± 1.4AFB 141.5 ± 2.5 142.9 ± 1.9

Potassium (mmol/L) (3 meq/L in dialysate) Post-HDF 5.8 ± 0.5 4.7 ± 0.4b,e

Pre-HDF 5.5 ± 0.2 4.7 ± 0.1b,e

AFB 5.2 ± 0.5 3.7 ± 0.3c

Potassium (mmol/L) (4 meq/L in dialysate) Post-HDF 5.3 ± 0.6 5.0 ± 0.1b,f

Pre-HDF 5.6 ± 0.3 4.9 ± 0.1b,f

AFB 5.5 ± 0.4 4.0 ± 0.2c

Chloride (mmol/L) Post-HDF 97.3 ± 5.8 98.6 ± 6.2Pre-HDF 97.2 ± 5.7 99.5 ± 7.0AFB 96.5 ± 7.4 97.0 ± 6.3

Ionized calcium (mmol/L) Post-HDF 1.20 ± 0.10 1.26 ± 0.11a

Pre-HDF 1.19 ± 0.09 1.27 ± 0.06b

AFB 1.15 ± 0.11 1.23 ± 0.07b

Total calcium (mmol/L) Post-HDF 2.33 ± 0.20 2.47 ± 0.21Pre-HDF 2.30 ± 0.28 2.45 ± 0.15AFB 2.35 ± 0.20 2.51 ± 0.16b

Bicarbonate (mmol/L) Post-HDF 22.0 ± 1.5 28.1 ± 1.1c,d

Pre-HDF 22.0 ± 2.2 27.1 ± 1.2c,f

AFB 22.8 ± 1.6 29.8 ± 1.9c

a p < 0.05.b p < 0.01.c p < 0.001, compared to beginning value.d p < 0.05.e p < 0.01.f p < 0.001, compared to AFB value.AFB: acetate-free biofiltration, HDF: hemodiafiltration.

F. DING ET AL.174

Artif Organs, Vol. 26, No. 2, 2002

inadequate quality of life and a substantially short-ened life expectancy compared to healthy subjects.Indeed, a series of reports has revealed that dialysis-related pathology, including b2M associated amy-loidosis, accelerated atherosclerosis, malnutrition,and infection, are still common features of long-termdialysis therapy.

In addition, problems related to acute hemodialy-sis, such as hypotensive episodes, cramps, vomiting,pruritus, and so on, still trouble both dialysis carestaffs and patients and call for more satisfactory, ef-fective dialysis techniques. Because a larger propor-tion of new patients entering renal replacementtherapy (RRT) are from so-called high risk patientgroups such as diabetics, the elderly, and those withcardiovascular complications or systemic diseases,the need for effective dialysis techniques is urgentand attractive.

In the past 20 years, 2 modern RRT modalities,namely online HDF and AFB, have become com-mercially available and appear to offer optimalforms to achieve this goal. Compared to conven-tional hemodialysis, a series of advantages for bothonline HDF and AFB can be observed: higher re-moval of small and larger molecular weight toxins

through the combination of diffusive and convectiveforces, increased hemodynamic stability found afterthe application of convective techniques, improvedintradialysis symptomatology and a subjective feel-ing of well-being in patients, better biocompatibilityprovided by high blood compatibility of the dialysismembrane coupled with ultrapure dialysate and ster-ile substitution fluids, optimal acid-base correction,and so on.

Both HDF and AFB are based on the same HDFtechnology. However, they do not share all featuresdue to differently designed technical characteristics.Because the literature lacks comparisons betweenthese 2 modalities, we carried out this prospectivecrossover study.

All 3 dialysis techniques delivered a satisfactoryclearance for low molecular weight toxins estimatedby Kt/V, corresponding to the guidelines of theAmerican National Kidney Foundation (DOQI)(Kt/Vsp 4 1.2) and recommendations by the HEMOstudy group (5) (Kt/Vsp 4 1.25 or Kt/Vdp 4 1.05) sothat a significantly higher value of Kt/V was achievedthan other patients not participating in the study inthe same center. Consistent with the results of ureaclearance, Kt/V delivered by online HDF was sig-

TABLE 5. Reduction rate during AFB, pre-HDF, and post-HDF

Post-HDF Pre-HDF AFB Difference

Urea 71.0 ± 7.9 70.7 ± 7.3 67.0 ± 6.5 NSCreatinine 59.7 ± 7.1 57.2 ± 6.7 57.4 ± 6.0 NSPhosphate 44.4 ± 16.4 42.0 ± 13.3 45.4 ± 13.2 NSb2-microglobulin (uncorrected) 66.2 ± 13.4 64.9 ± 13.6 61.2 ± 11.8 NSb2-microglobulin (corrected)a 70.1 ± 13.2 68.8 ± 13.4 65.0 ± 12.1 NS

a b2-microglobulin (corrected) 4 b2-microglobulin (uncorrected)/(1+ BW/0.2BWp)where BW is body weight change during dialysis and BWp is postdialysis bodyweight.

AFB: acetate-free biofiltration, HDF: hemodiafiltration.

FIG. 1. The graph shows effec-tive clearance in post-HDF, pre-HDF, and AFB (HDF: hemodia-filtration, AFB: acetate-freebiofiltration).

HEMODIAFILTRATION VERSUS BIOFILTRATION 175

Artif Organs, Vol. 26, No. 2, 2002

nificantly higher than that by AFB. It is well knownthat lower Kt/V is associated with increased mortal-ity and morbidity, but it is not clear whether a fur-ther increase in Kt/V (as in our onLine HDF study)to an already “normal” dose of detoxification (as inAFB) at the small molecular range has further ef-fects on mortality, an issue waiting for the final re-sults of the HEMO high dose studies (Kt/Vsp 4 1.6,Kt/Vdp 4 1.45) by the end of 2000. The differencebetween HDF and AFB in Kt/V could primarily beexplained by an increased convective contribution.However, it does not exclude a role of increaseddialysate flow rate in HDF. In this regard, Hauk et

al. (6) described a nearly 10% Kt/V dose rise whendialysate flow rate was increased from 500 to 800ml/min.

Compared to prestudy values, both HDF andAFB patients exhibited a tendency of decreasedphosphate and b2M by 10% to 20%. There is in-creasing evidence that uremic syndrome is causednot only by the accumulation of small, protein-bound water-soluble compounds but also by so-called middle-molecular-weight uremic toxins (7).Hyperphosphatemia has been implicated, either di-rectly or indirectly, in the pathogenesis of secondaryhyperparathyroidism and cardiac disease (8). An-

TABLE 6. Bone metabolism parameter changes in AFB, pre-HDF, and post-HDF

Baseline Start of crossover 4 weeks 8 weeks End of crossover

BAP (mmol/L) Post-HDF 4.03 ± 2.72 4.52 ± 3.71 4.45 ± 3.86 3.90 ± 3.14 3.86 ± 3.06Pre-HDF 4.31 ± 3.49 4.85 ± 4.11 4.31 ± 3.87 4.35 ± 3.84AFB 3.75 ± 2.47 3.95 ± 2.94 3.66 ± 2.70 4.20 ± 3.53

Osteocalcin (ng/ml) Post-HDF 75.3 ± 47.2 88.7 ± 67.6 106.1 ± 82.1a 97.8 ± 75.5 94.7 ± 71.2Pre-HDF 97.4 ± 75.3 100.1 ± 80.8 109.9 ± 80.4a 107.6 ± 92.0AFB 88.6 ± 67.9 97.4 ± 64.1a 86.2 ± 61.9 109.3 ± 78.3a

PYD (nmol/L) Post-HDF 325.5 ± 99.5 329.5 ± 135.9 227.3 ± 66.5b 293.6 ± 84.2 515.1 ± 278.6Pre-HDF 479.6 ± 131.0a 261.5 ± 77.6 333.3 ± 49.0 392.4 ± 117.5AFB 505.5 ± 285.2 307.5 ± 124.2 609.0 ± 392.5 473.8 ± 163.2a

DPYD (nmol/L) Post-HDF 57.5 ± 18.1 66.8 ± 28.9 30.9 ± 13.9b,c 55.9 ± 29.7 105.2 ± 51.0a

Pre-HDF 97.3 ± 36.0a 42.7 ± 16.1 58.9 ± 16.7d 83.5 ± 32.0b

AFB 88.4 ± 44.1 54.6 ± 22.5 88.4 ± 39.1a 111.2 ± 46.3a

iPTH (pg/ml) Post-HDF 1059.5 ± 827.3 926.5 ± 806.1 1123.9 ± 1063.7 895.0 ± 729.5 1172.9 ± 979.4Pre-HDF 1098.4 ± 1024.9 1071.8 ± 1111.5 1213.6 ± 1204.4 1245.6 ± 1256.6AFB 1232.9 ± 958.7 1001.4 ± 823.0 943.2 ± 739.9 1128.4 ± 1005.1

Aluminum (mg/ml) Post-HDF 27.6 ± 16.5 14.9 ± 15.9a 14.9 ± 9.9a 13.2 ± 13.8 10.6 ± 4.8c

Pre-HDF 13.3 ± 7.6a 12.9 ± 7.8a 14.1 ± 11.7 8.1 ± 5.3c

AFB 16.9 ± 16.6b 13.6 ± 9.8b 11.4 ± 5.4b 9.6 ± 5.7c

a p < 0.05.b p < 0.01.c p < 0.001, compared to baseline value.c p < 0.01, compared to AFB value.d p < 0.05.AFB: acetate-free biofiltration, HDF: hemodiafiltration, BAP: bone-specific alkaline phosphatase, PYD: pyridinoline; DPYD: deoxy-

pyridinoline, iPTH, intact parathormone.

FIG. 2. The graph showssingle-pool Kt/V and double-pool Kt/V in post-HDF, pre-HDF,and AFB (HDF: hemodiafiltra-tion, AFB: acetate-free biofiltra-tion).

F. DING ET AL.176

Artif Organs, Vol. 26, No. 2, 2002

other frequent complication in maintenance hemo-dialysis is dialysis-related amyloidosis with b2M con-sidered to be the major component of amyloidogenicdeposits. Preliminary in vivo studies (9) suggestedthat b2M is not an inert molecule but that it exertsadverse biological activity, especially at high concen-trations. In theory, it is reasonable to assume that apersistent reduction in serum phosphate and b2Mlevel might decrease or delay, if not prevent, somedialysis-related pathologies. Although our data donot permit confirmation of this point due to thesmall number of patients and the short observationalperiod, strong evidence has emerged in 10 years.

The predialysis b2M concentration in 3 groups wasmaintained at 25 mg/L by the end of the study, avalue numerically but not significantly lower thanthat at the beginning although higher than in theinvestigations of Canaud et al. (1) and Wizemann etal. (10). We speculate that several mechanisms mayaccount for this phenomenon. First, Wizemann’sprotocol, in which total membrane surface area of3.6 m2 and 60 L of substitution fluid per session was

used, was far more radical than our HDF practice,let alone AFB. Second, the observed period for eachcrossover was only 12 weeks whereas it has beenproposed that at least 9 months is needed to reach anew state of equilibrium (11). Third, nearly half ofour participants had been on post-HDF, and the ma-jority received high-flux dialyzer before entering thecurrent study, leading to a decreased possibility ofdetecting statistical difference. The difference inb2M clearance between HDF and AFB might beattributed most likely to different convective clear-ances but not to different dialysate flow rates, sug-gested by our previous research (4). Similar resultswere obtained by Hauk et al. (6), who reported thatincreasing dialysate flow rate from 500 to 800 ml/mindid not affect b2M removal ratio and clearance.

One of the primary reasons for designing thiscrossover study was to clarify whether or not AFB oronline HDF have different actions on bone metabo-lism parameters compared with conventional hemo-dialysis. Theoretically, both AFB and HDF may ex-ert a beneficial effect on bone metabolism due tobetter correction of acidosis, use of high biocompat-

FIG. 4. The graph shows albumin concentration changes in post-HDF, pre-HDF, and AFB (HDF: hemodiafiltration, AFB: acetate-free biofiltration).

FIG. 5. The graph shows albumin concentration changes duringthe whole study.

FIG. 3. The graph shows thepercentage of arterial bloodpressure changes during post-HDF, pre-HDF, and AFB (HDF:hemodiafiltration, AFB: acetate-free biofiltration).

HEMODIAFILTRATION VERSUS BIOFILTRATION 177

Artif Organs, Vol. 26, No. 2, 2002

ible membrane, sterile dialysate/infusion, increasedremoval of large molecular weight toxins, and so on.We did not detect any improvement in either boneformation or absorption markers by the end of studybut did detect a trend of deterioration in such pa-rameters as osteocalcin, PYD, and DPYD at severaltime points. We will discuss these findings under thefollowing aspects.

First, the current 12 week study period might betoo short to detect any change in bone markers. Inan 18 month prospective study, Lefebvre et al. (12)found that an optimal correction of metabolic acido-sis could prevent progression of secondary hyper-parathyroidism in hemodialyzed patients. Anothershorter investigation (but longer than ours) withpositive results was carried out by Vincenzi et al.(13). They reported a markedly decreased iPTHlevel on a 12 month AFB treatment. On the con-trary, in no study (14) shorter than 6 months werethere positive results.

Second, lower ionized calcium concentrations indialysate (1.25 in HDF and 1.5 in AFB in most cases)were used in our study, which might induce a nega-tive calcium balance during treatment. However,due to an oral CaCO3 medication for all patients, nototal negative calcium balance should be expected(15). Moreover, a stable predialysis or even an in-crease in postdialysis ionized calcium level could beachieved at the expense of calcium outflow frombone. In vivo data indicate that at least 1.75 mmol/Lof calcium in dialysate per infusion is indispensableto reach neutral calcium mass balance when per-forming AFB (13) or online HDF (16).

Third, as mentioned above, our study patientpopulation had been treated with an excellent he-modialysis characterized by the standard use of high-flux membranes and post-HDF in nearly half of theparticipants. It is thus reasonable to speculate thatthe chance of detecting differences from an ad-equately treated group of patients is minimal or thata much longer trial period is needed.

Last but not least, the current study was started atmidsummer after maximal exposure to sunlight andfinished at midspring after minimal sunlight. Recentdata (17) have shown that the winter season is one ofthe independent predictors for hypovitaminosis Dand that wintertime is associated with a higher inci-dence of vitamin D deficiency.

Data concerning periodic variation of circulatingbiochemical markers of bone remodeling in uremicpatients are still lacking, but it was proposed thatdefective skeletal mineralization was more commonin the United Kindom than in North America be-cause of, at least in part, less sunlight exposure in the

former area (18). Because no difference could befound in bone remodeling markers between onlineHDF and AFB at the end of study, we could at leastdraw such a conclusion that online HDF has a simi-lar effect as, if not better than, AFB on bone me-tabolism in the short term.

From a cardiovascular stability point of view,both AFB and online HDF can be considered to bethe most powerful tools for hemodynamically un-stable patients. Compared with a reported incidencefor hypotensive episodes between 20% to 30% perdialysis treatment, our patients’ incidence is fairlylow. The trend of more hypotensive episodes in AFBpatients, in our opinion, could be attributed to thefact that more patients during the AFB period hadan increased dry weight, which increases the chancefor hypovolumia at the end of the session in thesepatients.

All 3 treatment modalities share most of the im-portant features suggested to be responsible for im-proved cardiovascular stability: absence of backfil-tration, use of hemocompatible membrane,bicarbonate-buffered dialysate/infusate, higher con-vective clearance of vasoactive substances, and soon. When we examined intradialysis MBP change,AFB is associated with less drop of MBP duringtreatment than online HDF, this difference reachingsignificance between AFB and pre-HDF. The under-lying mechanisms for this phenomenon are not cer-tain yet. The impact of sodium and calcium concen-trations can be excluded because there were nosignificant differences between the 3 modalities.

One may guess that the discrepancy in cytokineproduction could be another relevant factor. Thisseems to be unlikely because our HDF patients arecompliant with strict hygiene handling. Further,Vaslaki et al. reported that the infusion of large vol-umes of online prepared substitution fluids and low-flux hemodialysis show a comparable cytokine pro-duction (EDTA Congress abstract, Madrid, Spain,1999).

Another possibility is the small amount of residualacetate in HDF dialysate per infusion. In this regard,acetate dialysis has been linked to a smaller increasein vascular resistance and a greater vasodilatationand myocardial depression, but it is not clear yetwhether the complete absence of acetate in AFB isresponsible for the stable blood pressure in our pa-tients. Recent in vitro data (19), however, clearlyindicated that even a small amount of acetate in themagnitude of HDF (2–4 mmol/L) is able to induceendothelial NOS activation and thus do harm tosome sensitive patients.

While all 3 groups of patients had comparable in-

F. DING ET AL.178

Artif Organs, Vol. 26, No. 2, 2002

tradialysis symptom scores, AFB patients seemed tohave higher interdialysis symptom scores, this differ-ence reaching statistical significance between AFBand post-HDF patients. Dialysis symptoms have amultifactorial pathogenesis that is not well definedyet. We strongly believe that differences in delivereddialysis dose and dialysate per infusion electrolytecomposition may contribute, at least in part, to dif-ferent symptom scores. Indeed, switching from HDFto AFB mode in some patients resulted in higherscores for itching both inter- and intradialysis, whichcould be solved partially by increasing calcium con-centration in dialysate.

One surprising discovery in our study was a sig-nificant drop in plasma albumin concentration, irre-spective of whether AFB or HDF was used. Serumalbumin level is not only an important index of nu-tritional status in end-stage renal disease patients butis also affected by many other nonnutritional factors,including external losses, decreased synthesis in theliver, and an increased uremic inflammatory re-sponse. Our patients received a very high dialysisdose for both small and middle molecule removal.They showed an unsuppressed appetite, reflected di-rectly by a very low score of anorexia in the ques-tionnaire and indirectly by an increased pretreat-ment potassium and stable urea level. This indicatesthat a possible low protein and caloric intake couldnot explain these findings. Also, there were no symp-toms and signs of augmented systemic and microin-flammation in these patients during the completestudy period. Convincing evidence (1,20) has shownthat both AFB and online HDF do not augment, butameliorate the degree of inflammation due to an im-proved overall biocompatibility and a better correc-tion of metabolic disorders such as acidosis. Hence,we strongly believe that the drop in serum albuminin the current study is most likely attributable to alarge amount of nutrient losses during dialysis.

With HDF techniques such as online HDF andAFB, a high transmembrane pressure (TMP) be-tween 300 to 500 mm Hg is not uncommon, espe-cially with post-HDF mode and/or near the end ofthe treatment session due to a progressively in-creased hematocrit. We recently found that a highTMP value is exponentially associated with a higheralbumin loss (unpublished data). Therefore, it is notsurprising that albumin levels of those patients re-ceiving hemodialysis before entering the studyshowed decreased albumin levels to a lesser extentthan HDF. Similar results also are provided by Ma-duell et al. (21). They discovered a reduction of al-bumin levels in the first month and of total protein inthe twelfth month after changing from conventional

HDF to online HDF. The large amount of proteinloss might be compensated partially by good appe-tite, resulting in a slow increase of serum albuminafter 8 weeks and a stable dry weight throughout thestudy. For this reason, long-term nutritional investi-gations and protein kinetic studies are needed toclarify this important issue.

A slight but significant drop in PO2 concomitantlywith an increase in PCO2, most apparent at 40 and 60min after the start of dialysis, was linked to patientstreated by online HDF but not to those by AFB. Thepathophysiology of hypoxemia has a multifactorialorigin and has not been completely understood.Among the supposed mechanisms, membrane-induced complement activation, the Bohr effectcaused by pH changes, reflex hypoventilation, andacetate-related effects are generally accepted. In thepresent study, AFB patients who received the sameF60S polysulfone dialyzer exhibited higher pH andlower PCO2 values than HDF patients, strongly sug-gesting that the occurence of hypoxemia duringHDF is mediated most likely by the small amount ofacetate, but not by other mechanisms. Indeed, ac-etate metabolism could cause an increased O2 con-sumption and a decreased CO2 production with aconsequent respiratory quotient reduction.

In addition, acetate is supposed to have a directdepressive action on the respiratory center and onthe myocardium, which results in an imbalance be-tween blood flow and ventilation of the lung andeventually the decrease of PO2. Despite a less than20% reduction of PO2 compared to predialysis valuesin our patients, the clinical relevance of such a re-duction in patients with a limited pulmonary andcardiac reserve remains to be established.

Consistent with the result from another prospec-tive study comparing AFB with HDF (22), the dif-ferences in biochemistry parameters among the 3modalities of treatment were not remarkable apartfrom better correction of metabolic acidosis andlower postdialysis potassium concentration in AFB.It should be emphasized that the modulation of theinfusion rate of the bicarbonate solution must befollowed with great care in the introductory phase ofAFB treatment. In fact, acid-base imbalance, moreoften metabolic alkalosis, is not rare even though therecommendation provided by the manufacturer isfollowed strictly. Santoro et al. (23) found a formulato predict the posttreatment bicarbonate concentra-tion that considers multiple factors including dialysistime, dialyzer type, blood flow, ultrafiltration rate,bicarbonate infusion rate, basal bicarbonate levelin blood, and hematocrit. Besides these, we sug-gested that vascular access recirculation should

HEMODIAFILTRATION VERSUS BIOFILTRATION 179

Artif Organs, Vol. 26, No. 2, 2002

be monitored in case of an unexpected postdialysisbicarbonate value due to decreased effective bloodflow. Not in keeping with the report of Maduell et al.(21), no impact on EPO dose and hematocrit werefound either between treatment modalities or duringthe research period. These could be explained by theshorter observational period in our study, which isinsufficient to detect small differences in EPO re-sponse.

In summary, our study demonstrates that both on-line HDF and AFB share most of the features ofoptimal renal replacement therapy. Online HDF issuperior to AFB in such aspects as an increased di-alysis dose both for small and larger molecularweight toxins and less interdialysis symptoms. Onthe other hand, AFB is associated with a smallerimpact on arterial blood gas and improved intradi-alysis hemodynamic tolerance. Some dialysis-relatedsymptoms and complications in the case of the AFBpractice could be attributable, at least in part, to thedialysate calcium level, recommending a higher cal-cium concentration in those cases.

Acknowledgments: We are indebted to the nurses andtechnicians at Dialyse-Gemeinschaft Nord e.V. for theirinvaluable help. We are also grateful to Dr. A. Trommerand Mrs. D. Freitag at Laborarztpraxis Dr. Muller, Ros-tock, for their expert laboratory help in measuring bloodsamples. Dr. Feng Ding is a nephrologist from HuashanHospital, Fudan University, Shanghai, PRC.

REFERENCES

1. Canaud B, Bosc JY, Leray H, Stec F, Argiles A, Leblanc M,Mion C. On-line haemodiafiltration: State of the art. NephrolDial Transplant 1998;13(Suppl 5):3–11.

2. Verzetti G, Navino C, Bolzani R, Galli G, Panzetta G. Ac-etate-free biofiltration versus bicarbonate haemodialysis inthe treatment of patients with diabetic nephropathy: A cross-over multicentric study. Nephrol Dial Transplant 1998;13:955–61.

3. James IT, Crowley C, Perrett D. Assay of pyridiniumcrosslinks in serum using narrow bore ion-paired reversedphase high performance liquid chromatography. J Chro-matogr 1993;612:41–8.

4. Ahrenholz P, Winkler RE, Ramlow W, Tiess M, Muller W.On-line hemodiafiltration with pre- and postdilution: A com-parison of efficacy. Int J Artif Organs 1997;20:81–90.

5. Eknoyan G, Levey AS, Beck GJ, Agodoa LY, Daugirdas JT,Kusek JW, Lewin NW, Schulman G. Hemodialysis (HEMO)study: Rationale for selection of interventions. Semin Dial1996;9:24–33.

6. Hauk M, Kuhlmann MK, Riegel W, Köhler H. In vivo effects

of dialysate flow rate on Kt/V in maintenance hemodialysispatients. Am J Kidney Dis 2000;35:105–11.

7. Vanholder R, De Smet R, Hsu C, Vogeleere P, Ringoir S.Uremic toxicity: The middle-molecule-weight hypothesis re-visited. Semin Dial 1994;14:205–18.

8. Amann K, Gross ML, London GM, Ritz E. Hyperphos-phataemia: A silent killer of patients with renal failure?Nephrol Dial Transplant 1999;14:2085–7.

9. Ferreira A, Urena P, Ang KS, Simon P, Morieux C, Souber-bielle JC, de Vernejoul MC, Drueke TB. Relationship be-tween serum beta 2-microglobulin, bone histology and dialysismembranes in uraemic patients. Nephrol Dial Transplant1995;10:1701–7.

10. Wizemann V, Lotz C, Techert F, Uthoff S. On-line haemo-diafiltration versus low-flux haemodialysis: A prospective ran-domized study. Nephrol Dial Transplant 2000;15(Suppl 1):43–8.

11. Raj DSC, Ouwendyk M, Francoeur R, Pieratos A. b2-microglobulin kinetics in nocturnal haemodialysis. NephrolDial Transplant 2000;15:58–64.

12. Lefebvre A, de Vernejoul MC, Gueris J, Goldfarb B, GrauletAM, Morieux C. Optimal correction of acidosis changes pro-gression of dialysis osteodystrophy. Kidney Int 1989;36:1112–8.

13. de Vincenzi A, Bellazzi R, Santagostino M, Romanini D, NaiM, Gazo A, Bacchella L, Gini A. Calcium mass balance andbehavior of intact immunoreactive parathyroid hormone inacetate-free biofiltration: acute and one-year evaluation.Blood Purif 1994;12:85–94.

14. van der Meer AMS, ter Wee PM, Donker AJM, van DorpWT. Dialysis efficacy during acetate-free biofiltration.Nephrol Dial Transplant 1998;13:370–4.

15. Ahrenholz P, Winkler RE, Ramlow W, Tiess M, Thews O.On-line hemodiafiltration with pre- and postdilution: Impacton the acid-base status. Int J Artif Organs 1998;21:321–7.

16. Malberti F, Corradi B, Tetta C, Imbasciati E. Calcium balanceand serum calcium fluctuations in on-line haemodiafiltrationin relation to ultrafiltration rate and dialysate calcium con-centration. Nephrol Dial Transplant 1994;9:1759–64.

17. McKenna MJ. Differences in vitamin D status between coun-tries in young adults and the elderly. Am J Med 1992;93:69–77.

18. Llach F. Renal osteodystrophy. In: Jacobs C, et al., eds. Re-placement of Renal Function by Dialysis. Kluwer, 1995:1167–70.

19. Amore A, Cirina P, Mitola S, Peruzzi L, Bonaudo R, Giano-glio B, Coppo R. Acetate intolerance is mediated by en-hanced synthesis of nitric oxide (NO) by endothelial cells. JAm Soc Nephrol 1997;8:1431–6.

20. Galli G, Panzetta G. Acetate free biofiltration (AFB): Fromtheory to clinical results. Clin Nephrol. 1998;50:28–37.

21. Maduell F, del Pozo C, Garcia H, Sanchez L, Hdez-Jaras J,Albero MD, Calvo C, Torregrosa I, Navarro V. Change fromconventional haemodialfiltration to on-line haemodiafiltra-tion. Nephrol Dial Transplant 1999;14:1202–7.

22. Movilli E, Camerini C, Zein H, D’Avolio G, Sandrini M,Strada A, Maiorca R. A prospective comparison of bicarbon-ate dialysis, hemodiafiltration, and acetate-free biofiltrationin the elderly. Am J Kidney Dis 1996;27:541–7.

23. Santoro A, Spongano M, Ferrari G, Bolzani R, Augella F,Borghi M, Briganti M, Cagnoli L, Docci D, Feletti C, et al.Analysis of the factors influencing bicarbonate balance duringacetate-free biofiltration. Kidney Int 1993;41(Suppl):S184–7.

F. DING ET AL.180

Artif Organs, Vol. 26, No. 2, 2002