Influence of Online Hemodiafiltration on Hemoglobin Level, ESA-Dosage and Serum Albumin – A Retrospective, Multicenter Analysis

In renal replacement therapy (RRT) a wide range of uremic toxins have to be removed (Vanholder et al., 2003; Vanholder et al., 2008). It is well known that the combination of diffusive and convective dialysis strategies (online hemodiafiltration, olHDF) improves the removal of uremic toxins, i.e. middle molecules, hydrophobic substances and protein (albumin) bound materials (Krieter et al., 2005; Ahrenholz et al., 2004; Ronco et al., 1999; Kim, 1994; Testa et al., 2006; Meyer et al., 2005; Mandolfo et al., 2006; Kanter et al., 2008). In the presence of ultrapure dialysis fluid which is ultrafiltrated by endotoxin restraint systems (Weber et al., 2000; Canaud et al., 2001; Minetti et al., 1985; Hakim et al., 1984) and biocompatible high flux dialysis membranes the convective diffusive treatment significantly prevents the complement activation (Braun et al., 1995; Savica et al., 2006; Jorres et al, 1999; Horl et al., 1986) in the first minutes of olHDF session and removes proinflammatory substances (cytokines) too (Bellomo et al., 1991; Filiopoulos et al., 2008; Haas et al., 2007; Libetta et al., 2007; Mariano et al., 2005). The intra and interdialytic inflammations are reduced (Ramirez et al., 2007; Ramirez et al., 2007; Carracedo et al., 2006). The suppressed inflammatory process during olHDF leads to an increasing serum albumin concentration via an increased synthesis rate. In spite of a varying albumin removal caused by different dialysis membranes with/without adsorptive character and pore size (Pichaiwong et al., 2006; Yamashita, 2007; Tomo et al., 2008; Winchester et al.,2003; Winchester et al., 2004) the albumin synthesis rate increases by absent inflammation (Giordano et al., 2001). It is shown that in olHDF the ESA dosage needed to reach the hemoglobin goal is reduced (Vaslaki et al., 2006; Bonforte et al., 2002; Eiselt et al., 2000). The efficacy of dialysis measured by single pool Kt/V could be improved (Ahrenholz et al., 1997; Ding et al., 2002). There is evidence of longer survival of patients treated by olHDF versus hemodialysis (HD) independently of dialysis dosage (Canaud et al., 2008; Panicchi et al., 2008). The cycling of hemoglobin levels depends on inflammatory episodes and malnutrition (Del et al., 2005; Brimble et al., 2005). This retrospective, non randomized, multicentre, descriptive clinical evaluation examined the influence of olHDF on hemoglobin concentration (Hb), ESA dosage (ESA), Hb variability (Hbvar), albumin and CRP.


Materials and methods
233 chronic hemodialysis patients were included in this clinical evaluation (dialysis center 1 (D1) n= 94, D2 n= 35, D3 n= 104 patients). 54.9% were male; the mean age was 63.8 years (range 22 -89). The patients in all three centers were comparable with regard to gender distribution, mean age, mean time on dialysis, and distribution of underlying kidney disease.
The clinical evaluation was carried out for 12 months retrospectively. Laborchemical parameters were estimated for hemoglobin (Hb) every 2 weeks (labanalyzer), CRP (turbidometry), albumin (alb; nephelometry) and ferritin (chemiluminescence technique) every three months. Serum iron (photometry) and transferrin (turbidometry) were necessary to calculate the transferrin saturation (every 4 weeks). Single pool kt/V was evaluated every 3 months with the Daugirdas technique (Daugirdas, 1993). Intraindividual variability of hemoglobin (Hbvar) was defined as the difference between minimal and maximal concentration (range) and by time to reach the target between Hb 6.8 mmol/l and 8.0 mmol/l within 9 months. Relevant changes in ESA dosages were defined as an elevation greater than two fold and lowering of a half of the ESA dosage, the end of ESA application or the start with more than 4200 U/week. Hemodialysis (HD) was performed by MTS 5008 (Fresenius Medical Care), low flux dialyser FX8, FX10 (helixone, Fresenius Medical Care), Q B 300 ml/min, Q D 500 ml/min, ultrapure dialysis fluid, online hemodiafiltration (olHDF) by MTS 5008 (Fresenius Medical Care; automatic procedure with factor 1.2), high flux dialyser FX 60, FX80 (helixone, Fresenius Medical Care), Q B 300 ml/min, Q D 350…..360 ml/min, Q S 51….60 ml/min, ultrapure dialysis fluid and Nikkiso DBB 05 (Nikkiso Medical Ltd.), high flux dialyser FDY 15 G (PEPA, Nikkiso Medical Ltd.), Q B 300 ml/min, Q D 700 ml/min, Q S 60 ml/min, ultrapure dialysis fluid (Q B … blood flow; Q D …dialysate flow; Q s …substitution flow). The group "mixed" contained patients started with HD and switched to olHDF (at least 6 months olHDF). We compared the mean values of collected serum parameters three times a month. Descriptive statistical evaluation was calculated by mean, standard deviation and significance by Wilcoxon test, correlation by Spearman rang correlation. The level of significance was defined as p< 0.05.

Results
The distribution of ESA applications in the three observed dialysis centers can be seen in table 1. Totally 185 of 233 patients received at least one ESA dosage. The mean value of ferritin was 538 mg/L. The transferrin saturation (TSAT) did not differ significantly in the observed dialysis units. In nearly all patients (98.9%) an adjustment of ESA dosage was essential. Relevant changes in ESA dosages were defined as an elevation greater than two fold and lowering of a half of the ESA dosage, the end of ESA application or the start with more than 4200 U/week. The mean value of Hb (measured per patient over the whole study time) was 7.35 mmol/l (Table 3). In patients without ESA application during the 12 months study the mean value of Hb was larger (7.66 mmol/l) in comparison to patients with ESA dosage (7.27 mmol/l).

Discussion
Our retrospective analysis was performed in three different dialysis centers for 12 months. The D1 center had the largest percentage of patients treated with olHDF (olHDF+"mixed") (57 % in D1 vs. 9 % in D2, 29 % in D3). In D1 the lowest dosage of ESA to reach the Hb target was used (  Table 11) and a negative one with the mean ESA consumption (p = 0.05). These results comply with the expectation because an improved Hb value is connected with a larger TSAT level and reduced ESA needs. The treatment efficacy (single pool and equilibrated Kt/V; spKt/V, eKt/V), which was measured periodically in the 3 dialysis units, did not show any significant influence on ESA www.intechopen.com dosage and Hb levels. But the subanalysis calculating the impact of the different treatment modes on Kt/V resulted in a significant increased spKt/V for olHDF treatments compared with HD (1.62±0.28 for olHDF versus 1.48±0.44 for HD; Table 8).
Interestingly the correlation analysis also shows a highly significant positive correlation of the mean albumin level with the mean Hb values (p < 0.001, Table 11) and a negative one with the mean ESA dosage (p = 0.0003). Simultaneously CRP is negatively correlated with Hb (p = 0.0004, Table 11). The significant difference in albumin concentration most likely played the decisive role for ESA dosage and Hb level (Ward, 2005). It is known that in patients who underwent convective-diffusive treatment the ESA dosage could be reduced (Vaslaki et al., 2006;Bonforte et al., 2002;Eiselt et al., 2000). That observation was confirmed by our results, reaching an economically interesting level of savings in ESA costs: Fig. 1. Typically, convective diffusive procedures are characterized by an additional removal of hydrophobic middle molecules and protein (albumin) bound uremic toxins depending on the membrane characteristics (hydrophobic areas, pore size, adsorptive properties, biocompatibility) (Ahrenholz et al., 2004;Panicchi et al., 2008). The loss of protein bound substances leads to a membrane determined l o s s o f a l b u m i n d u ring olHDF sessions (Ahrenholz et al., 2004;Samtleben et al., 2003;Combarnous et al., 2002). In low flux dialysis protein removal only occurs with adsorptive membranes (PMMA, polyacrilonitrile) with decreasing dialysis efficacy for water soluble toxins (Parzer et al., 1993). This removal of albumin can be compensated after a time of about 12 weeks in the absence of relevant inflammation (Ding et al., 2002;Kaysen et al., 1997). In chronic ambulant peritoneal dialysis protein losses are in-between 6 to 10 g/d and albumin losses up to 5 g/d over the peritoneal membrane (Kaysen et al., 1984). Albumin losses during renal replacement procedures are generally thought of being unwanted, as low serum albumin correlates with poor outcome in dialysis patients. Therefore, olHDF, that technically spoken is an albumin-loosing therapy, might carry the danger of exposing the treated patients to threads associated with low albumin levels.

Mean Albumin HD vs olHDF
It is striking that in our analysis the olHDF group had the largest serum albumin concentration (Fig. 2, Table 12). All patients of the "mixed" group (containing patients that had switched from HD to olHDF) showed an increase in albumin level rather than a decrease. Moreover, olHDF can remove proinflammatory substances such as cytokines (Bellomo et al., 1991;Lee et al., 2004). Again, we could confirm this phenomenon with lower CRP levels in the olHDF group vs. HD group (9.96+/-8.28 mgl/l vs. HD 16.07+/-16.26 mg/l, p=0.02), see Fig. 3, and Because albumin is a negative acute phase protein we can, in general, expect higher concentrations at lower inflammation (Panicchi et al., 2006;Kaysen et al., 1997). However, none-biocompatible membranes and partly low flux hemodialysis increases proinflammatory cytokines such as TNF-alpha. Ultrapure dialysis fluid is of relevant importance to prevent inflammation (Panicchi et al., 2008). On the other hand complement activation plays a role for inflammation during the dialysis sessions therefore biocompatible membranes are urgently necessary (Hakim et al., 1984). In the olHDF method as use in this study, both ultrapure dialysate and biocompatible membrane materials were used, enabling clear attenuation of procedure-associated inflammatory processes. This attenuation of inflammation to us seems the key factor for increased albumin production that even makes up for procedure-associated albumin losses. The nutritional situation (nPCR) has only a secondary influence (Savica et al., 2006;Stenvinkel, 2005). Hbvar also depends on inflammation and albumin concentration (Brimble et al., 2007). Hbvar in olHDF is lower than in HD because of less inflammation and higher concentration of albumin.

Conclusions
In a retrospective, descriptive, multicentre study the influence of olHDF on Hb Level, ESA dosage and Hbvar was evaluated. 233 patients were included in the clinical analysis in three dialysis departments (D1 n=94; D2 n= 35, D3 n= 104). Mean dialysis efficacy expressed as spkt/V by Daugirdas was comparable in all dialysis units. We found differences in the frequency of olHDF in the dialysis departments followed by varying parameters of inflammation (CRP) and nutrition (albumin). It can be demonstrated that patients who underwent olHDF showed the highest serum albumin levels and the lowest signs of inflammation (CRP). This combination leads to significantly higher Hb concentrations and surprisingly lower ESA dosages to reach the target Hb in ol HDF vs HD. Due to the reduced inflammation Hbvar was improved in olHDF vs HD. There is a correlation between serum albumin concentration, Hb level and ESA dosage. OlHDF could be the gold standard for prevention of inflammation because of removal of proinflammatory substances and hydrophobic and protein bound uremic toxins. OlHDF influences positively inflammation, nutrition, Hb level, Hb variability and required ESA dosage in chronic renal replacement therapy.