Patients

The combined observational study included data from 60 HD sessions performed by 20 (8 women and 12 men) long-term chronic HD patients from the same center (Umea University Hospital, Sweden). All patients were in stable condition without a recent history of infections, embolic events, or active tumors within the latest three months. Their primary renal diagnoses: undefined chronic renal failure (n =7), polycystic kidney disease (n =3), diabetic nephropathy (n =3), glomerulo-nephritis (n =3), pyelonephritis (n =2), and hypertension (n =2). The mean age was 65 years (±10). Women did not differ from men in age (65 ± 12 vs 66 ± 10 years) or vintage time of HD (33 ± 28 vs 43 ± 27 months). The vascular accesses used were central dialysis catheters (CDC, 30 sessions), arteriovenous fistula (AVF, 21 sessions), and AV-graft (AVG, 9 sessions). The mean blood pump flow rate used was 317 ± 45 mLmin-1. Additional predialysis data are presented in Tables 1 and 2 and in Supplement Tables 1, 2 and 3. The patients were described in more detail in a previous study that investigated their cardiac condition [24].

Table 1 The mean, standard deviation (Std. Dev), median, minimum (min) and maximum (max) values of the predialysis variables months being HD patient (vintage), estimated normal weight (Dry weight), weight gain between dialyses – estimated as interdialytic weight gain (IDWG), low molecular weight heparin dose (LMWH) as Units/patient and Units/kg body weight (bow), duration of HD session, removed fluid by ultrafiltration (UF), and speed of removed fluid in relation to IDWG/hour of HD (UF-ratio)Table 2 Distribution of D-dimer values given as the mean (±SD) and median (min-max) at predialysis (0min), at 30min, at 180min, and the differences (δ) between two measures. N.S. = not significant

The local ethics committee in Umea, Sweden, had approved the research protocol (EPN 05-138M, addition date 20071016; EPN2012-42-31M, 20120306). Patients were informed and consented to participate in the study that complied with the Declaration of Helsinki.

Dialyses were done using Fresenius 4008, Fresenius 5008, and Gambro AK200 monitors. LMWH anticoagulation, as bolus dose at start of HD, was given as tinzaparin in 57 HDs and as dalteparin (due to adverse events by tinzaparin) in three HDs.

Study design

Each patient performed three comparative study dialyses as mid-week sessions. The other dialyses, between the study dialyses, were performed with standard HD or hemodiafiltration (HDF) using various Fresenius dialyzers. All study dialyses were performed with low-flux dialyzers made of polysulfone and of the same size (1.8 sqm), either steam or gamma sterilized and using either a high or low blood level in the venous chamber (air trap).

Option 1: a dry-stored low-flux dialyzer (F8HPS, Fresenius Medical Care, steam sterilized) with a low blood level in the venous chamber, high enough not to induce alarm (Mode-DL)

Option 2: a dry stored low-flux F8HPS dialyzer with a high blood level (Mode-DH).

Option 3: a wet-stored low-flux dialyzer (Rexeed18L, Asahi Kasei Medical, gamma sterilized) with a high blood level (Mode-WH).

All patients were dialyzed by all three modes in a cross-over and randomized order. Low-flux membranes were used to minimize loss of middle molecular or larger size of laboratory markers by clearance through the membrane. Blood samples were drawn before HD and at 30min and 180min of HD. During HD, all samples were collected from the arterial site of the dialysis circuit before entrance into the dialyzer. Blood was collected in citrate and EDTA tubes, centrifuged and plasma stored at -80°C until analysis.

Values during dialysis were adjusted for the change in plasma concentration and erythrocyte volume fraction caused by infusion and/or ultrafiltration in accordance with Schneditz et al [25]. Fluid retention between dialyses was estimated as inter-dialytic weight gain (IDWG) calculated as the percentage increase of body weight between two dialyses [26, 27].

Laboratory analyses performed within 2 hours of blood collection were hemoglobin (Hb, ref 100-120 gL-1), erythrocyte volume fraction (ref 0.30-0.36), neutrophil granulocytes (Neutr, ref 1.8-6.3 x 10E9L-1), lymphocytes (Lymph, ref 1.0-3.5 x 10E9L-1), monocytes (Monoc, ref 0.3-1.2 x 10E9L-1), eosinophils (Eos, ref 0.07-0.3 x 10E9L-1), and thrombocytes particle concentration (TPC, ref 145-348 x 10E9L-1). Remaining analyses were performed later (samples stored at -80°C). A specific variable was analyzed at the same time to avoid errors between series. Variables were NT-pro-BNP (ProBNP, ref value <125 ngL-1 for patients aged 0-74 and <450 ngL-1 for elderly, Roche) and highly sensitive Troponin T (Troponin, ref <15ngL-1, Roche). ProBNP above the upper limit of 70,000 ngL-1 (of the laboratory) was set as 70,000 ngL-1. For these series (those above 70,000), no differences in ProBNP value between predialysis and the 180min data during HD could be calculated.

Other variables were pentraxin 3 (PTX, Perseus Proteomics Inc., Tokyo, Japan, ref <3.5 µg L-1), activated complement factor 3 (C3a, ref 20-130 µg L-1, according to Ekdahl et al. [28]), thrombin antithrombin complex (TAT, ref <3µg L-1, Enzyme Research Laboratories Ltd., Swansea, UK), for C3(H2O) i.e., full length complement factor 3 (iC3) with a broken thiol ester, analyzed according to Ekdahl et al. [28] with a tentative reference value of 95-1015 µL-1 based on 20 healthy individuals, C-reactive protein (CRP, ref <5 mg L-1, Immulite 1000, Siemens Healthcare, Erlangen, Germany), von Willebrand factor (vWF, ref 206-238 U L-1, Instrumentation Laboratory), tissue plasminogen activator activity (tPAact, ref 0.2-2 IUmL-1, Chromolize tPA activity, Biopool, Umeå, Sweden), tissue plasminogen activator mass (tPAm, ref 1-20 µgL-1, Trinity Biotech, Ireland Limited, IDA Business Park, Wicklow, Ireland), plasminogen activator inhibitor 1 mass (PAIm, ref 4-43 µgL-1, Tcoag, Trinity Biotech, Ireland Limited, IDA Business Park, Wicklow, Ireland), and fibrin degradation product D-dimer (D-dimer, ref <0.20 mg L-1, ACL TOP 700 LAS, 750 LAS, Instrumentation Laboratory).

IDWG was calculated as the percentage of weight gain between two dialyses in relation to the dry body weight of the patient. The removed fluid during HD was defined as ultrafiltration (UF, mL). The UF-rate was defined as the speed of removed fluid given as a rate of the IDWG%/hour of HD, such that if the IDWG was 4% of the body weight a 4-hour dialysis with continuous removal of fluid would result in a removal of 1% of body weight, i.e., 1 IDWG/hour. This would give a UF-rate of 1 using a longitudinal elimination.

Statistical analysis

Calculations were done for numerical data and differences in concentrations (delta-values) present at predialysis (before start of HD), at 30min, and 180min of HD. With a two tailed p value and alfa error of <0.05 and power of 0.80 nineteen pairs are sufficient.

Values were compared as pairs, groups, and aggregated.

Data were analyzed using SPSS (PASW Statistics for Windows, Version 28. Chicago: SPSS Inc.). The Wilcoxon signed rank test was used for paired samples, and Mann-Whitney U test used for independent samples. Correlations were analyzed with the non-parametric Spearman’s test (rho-value). Values are expressed as median (minimal and maximal) and mean (± standard deviation, SD). Significant values were defined as a two-tailed P-value <0.05. The P-value should be put in relation to the extent of secondary analyses, considering the Bonferroni concept of alfa-error for multiple analyses. Since every patient performed all three modes of treatments, the analyses were performed for all aggregate dialyses (n =60) and for each specific mode of HD (each n =20). Multiple linear regression analyses with D-dimer as dependent factor were performed using the stepwise model, including the variables age and significant variables from the bivariate analysis.