Online-haemodiafiltration vs. conventional haemodialysis: a cross-over study

In this cross-over study comparing 3 consecutive of 6-month periods (HDF1 − HD − HDF2), we found that HDF protocols were associated with lower albumin and ß2-m serum levels. No others clinical or biological parameters were found to differ during the 3 periods. A convective volume < 21 L was associated with the lower albumin levels during HDF.

In a 4-years observational study in 2006, slight decreases in serum albumin and prealbumin levels were reported during the first 6 months following a switch from HD to HDF with increased levels observed thereafter [17]. Ok et al. also reported lower albumin levels in the low-efficiency HDF arm of their prospective Turkish study [3]. Decreased serum albumin levels during the first 6 months have also been reported after switching from high-flux HD to HDF [18]. However, these observations were not confirmed in a study in Balkan countries [19], and Movilli et al. reported the absence of an albumin level decrease during an HDF protocol [9].

The lower serum albumin level observed during the HDF periods in the present study could be due to a dialysate albumin loss, as reported previously by Combarnous et al., who observed an albumin loss of 1000–6800 mg/session [15]. In 2004, Ahrenholz et al. reported a total albumin loss of 300–7000 mg/session, depending on the type of dialyzer used [14]. Unfortunately, we did not measure the albumin loss in the dialysate. In the CONTRAST study, de Hoedt et al. reported no difference in the rate of change in albumin between the HDF and low-flux HD arms [20]. However, only annual data was reported, and short-term evolution could have been missed. As serum albumin decreased mainly in patients with lower convection volume, the relationship between the change in albumin levels and the HDF technique itself remains unclear.

A decrease in ß2-m serum values during HDF was reported by Zehnder et al., who demonstrated its absorption to the polysulfone membrane [10]. Others studies have also reported a significant decrease in ß2-m levels after switching patients from HD to HDF [9, 21, 22]. However, in the Turkish study, there were not differences between the 3 arms involving HDF and low- and high-flux HD [3]. In the CONTRAST study, ß2-m decreased significantly in the HDF arm, especially in cases when residual kidney function was low [2]. Lastly, in the ESHOL study, ß2-m levels were similar between the HD and HDF arm [4]. A role for residual renal function has been hypothesized to account for this non-expected evolution. However, the factors associated with ß2-m decline during HDF have not been clearly reported. Our observation that diabetes was associated with smaller serum ß2-m decline requires further confirmation and explanation.

The impact of serum ß2-m on the survival in haemodialysis patients is controversial [23, 24] and depends on the patient’s nutritional factors and diabetic status; however, the negative association of low albumin levels with poor outcomes is clear [25]. Thus, a negative impact of HDF on serum albumin would be expected to associate with poor outcomes, but this effect has not been reported in the literature. Even if the question of a survival advantage for HDF is still under debate, the data do not confirm any disadvantage for HDF compared with HD.

HDF was not observed to have any impact on mean phosphate level or on phosphate binder requirements in our study. The literature on this subject remains controversial. In 1999, Zenhder et al. reported an increase in phosphate clearance during HDF vs. high-flux HD [10]. In 2010, another group reported lower serum phosphate levels in HDF-treated patients, but data on protein intake or phosphate binders were not provided [26]. Lornoy et al. reported higher phosphate removal during HDF when compared with HD, especially in the low-normal phosphate range, but not in cases where phosphate levels were high [27]. Others studies have also reported decreased phosphate levels during HDF vs. HD [9, 28‐30], mostly using higher blood flow rates. However, in 1991, Man et al. reported no advantage of HDF for phosphate mass transfer [23], and Ok et al. reported no differences in phosphate levels between HD and HDF groups [3]. Our observation confirms that the impact of HDF on phosphate levels is not significant. Moreover, we did not observed any changes in serum levels of PTH or bone markers, unlike a previous study that demonstrated, reductions in both PTH and bone alkaline phosphatase after switching from HD to HDF [31]. However, calcium mass transfer, a potentially key point, was not evaluated as in our study. Regarding small molecule clearance, we did not find differences between periods similar to Zehnder et al. who reported no advantage of HDF for urea and creatinine clearance [10]. By contrast, Movilli et al. [9] and the CONTRAST study [2] reported an increase in dialysis dose when using HDF vs. low-flux HD, but higher blood flow rates in the HDF arm may have biased the results.

We could not confirm any hemodynamic advantage of HDF, as we did not observe any difference in the number of hypotensive episodes or changes in blood pressure values during the 3 phases of the study. These finding are not consistent with results from postdilution HDF [5] and predilution HDF [32] studies. Additionally, the ESHOL study reported fewer hypotensive episodes in the HDF arm [4], especially when higher convective volumes were achieved, as reported by Mora-Bravo et al. [6]. However, the absence of a haemodynamic advantage for HDF has also been reported under strictly controlled conditions in 12 stable patients [33]. Similarly, no difference in blood pressure was found for HDF in association with intracellular or extracellular volume changes during sessions [34]. The favourable impact of HDF on haemodynamic stability is hypothesized to be due to higher sodium mass transfer in some cases, even if this phenomenon has been poorly documented.

We could not confirm better control of anaemia during HDF, as reported by Vilar et al. in a large observational study [35], or the correction of anaemia with reduced dose of ASE observed 9 months after switching 32 patients to HDF [11]. We also did not find any significant changes in CRP levels during the 3 periods in contrast to the results from a large multicenter crossover study [36] and 2 observational studies [5, 12].

Our study has numerous limitations including that the initial dialysis technique was not randomized, the size of the studied cohort was small, and we did not measure albumin loss in the dialysate. However, the cross-over design of the study could give a more sound comparison of the 2 modalities. Nevertheless, there was also some drop out and the patients’ state may have changed after 6 or 12 months.