Online haemodiafiltration (OLHDF) is the most complete haemodialysis technique currently available, as it is capable of removing significant quantities of low-, medium- and high-molecular weight uraemic toxins, in direct relation with the convective transport volume obtained.1 Post-dilution OLHDF is the most used form and has proven efficacy. It is known to be a safe technique which improves intradialytic haemodynamic tolerance and increases survival.2–4

OLHDF uses high-flux biocompatible dialysers, providing greater clearance per unit of surface by combining the processes of diffusion and convection. To date, the membranes used in this technique have been synthetic. In the study by Maduell et al.,5 in which multiple dialysers for OLHDF were compared, they considered that cellulose triacetate had a low purification of β2-microglobulin and limited its use in OLHDF due to high transmembrane pressure (TMP). An asymmetric cellulose triacetate membrane (ATA®) was recently marketed in the Solacea® (Nipro) dialyser with a KUF of 72ml/mmHg/h/m2 and configured in order to perform OLHDF. In accordance with the manufacturer's data, the complement activation produced is similar to that of the synthetic membranes, demonstrating its biocompatibility. Its long-term biocompatibility through its effect on monocytes and interleukins (IL) is still to be determined. As yet, there is no publication on the convective volume (Vconv) obtained, the adaptation to different OLHDF systems, the clearance capacity for different types of molecules or biocompatibility in daily clinical practice. The recent description of various cases of hypersensitivity with biocompatible membranes, such as polysulfone,6 in which cellulose triacetate is used as an alternative, has led to even greater interest in the study of this type of membrane.

To describe the features and the in vivo behaviour of the ATA® membrane in order to identify: its clearance efficacy, biocompatibility and ease of use in clinical practice.

This was a prospective, observational study in three hospital haemodialysis units (Hospital de La Princesa, Hospital Príncipe de Asturias and Hospital Infanta Leonor, Autonomous Community of Madrid, Spain) in which the normal synthetic dialyser that each patient had for OLHDF was replaced with an ATA® dialyser, with all other parameters remaining the same. The study (LIB 09/2015) was reviewed and approved by the Hospital Universitario Príncipe de Asturias IEC (Alcalá de Henares, Madrid).

The study included 23 patients (15 males, 8 females) with a median age of 65 (41–85). The causes of kidney disease were: diabetic nephropathy in seven cases, glomerular disease in six cases, nephroangiosclerosis in two cases, reduced kidney mass in two cases, hepatorenal polycystic disease in two cases, chronic pyelonephritis in one case and unknown origin in three cases. The median time on renal replacement therapy was 98 (17–315) months. Sixteen patients had arteriovenous fistula and seven had a catheter. The mean dry weight was 73.4 (16.7) kg.

A total of 287 sessions were performed, and 264 complete sessions were collected (sessions excluded were those in which the effective time differed by more than 15min from the scheduled time, or if the monitor had been changed). One patient was lost to follow-up due to vascular disease requiring hospital admission .

The results of dialysis parameters and molecule clearance are shown in Tables 2 and 3, respectively. The FF obtained with the different monitors was: 5008® 28.9 (3.9)%, AK 200® 31.8 (1.8)%, Artis® 32.6 (4)% and DBB-07® 27%. There was a negative correlation between the initial Hb and the Vconv (r=−0.5, p>.01) and the Kt (r=−0.5, p<0.01). There were no complications reported and there were no warnings issued by the dialyser and it was not necessary to make changes to the Ultracontrol® system. As coagulation problems did not occur the dose of heparin was not changed.

Finally, Table 4 lists the acute biocompatibility data while Table 5 shows details on inflammation parameters. Both tables show that there is no complement activation when using a cellulose membrane, and there is not worsening in inflammatory molecules after one month of treatment. There was no significant correlation between CRP and the other inflammatory parameters studied.

Our results show that the membrane studied has the necessary characteristics to perform OLHDF; it obtains a suitable Vconv, Kt and RR for different substances, it does not cause technical problems and nursing work is not affected beyond daily clinical practice. Also, the membrane is biocompatible (it does not activate the complement in the short-term or induce changes in the inflammatory profile in the medium term, based on the data obtained on activation of IL and monocytes).

The two clinical parameters used on a day-to-day basis to assess OLHDF's suitability are Kt and Vconv.9,10 The mean Kt reached was 56l, this value is greater than the target suggested in the guidelines, and, although the optimal Vconv remains to be decided, a recent study11 showed that increasing Vconv improves survival linearly between ∼55 and 75l/week; our results indicate that the maximum volume was achieved. The clearance capacity obtained as assessed by RRs for substances of different sizes were also similar to those presented in other studies which feature high-permeability dialysers to perform OLHDF,12,13 taking into account the differences which may be found depending on the dialysis time and the surface used.

According to the EUDIAL group, the dialyser used to perform OLHDF must have the following features: high flux, ultrafiltration coefficient greater than 20ml/mmHg/h/m2, permeability coefficient for β2-microglobulin greater than 0.6, and obtain an effective convective transport greater than 20% of the total processed blood.7 Given that each monitor uses different methods to optimise the FF,14 the results obtained were different depending on which monitor we used, but the results were greater than 27% in all monitors. The classic description of cellulose membranes is that they are thinner than synthetic membranes and have a capillary wall that is uniform and symmetric in composition, without having the necessary resistance to support the high pressures generated in OLHDF.15 However, the ATA® membrane is asymmetrical, which allows to have a low TMP and, therefore, to perform OLHDF achieving adequate clearance, in accordance with the measurement parameters used.

Regarding clinical aspects, there were no significant problems in the daily practice, we observed a well adaptation to the different systems which the monitors use to achieve an adequate Vconv, as reflected in the different FFs obtained. This is important fact, since the absence of alarm warnings improves the nurse's work load and means that the dialyser adapts well to the technique.

The idea persists that the contact of these cellulose membranes with blood causes a major inflammatory response, classifying them as non-biocompatible membranes. This inflammatory response could be acute, with complement activation and leukopaenia after 30min of the session, or have consequences in the longer term through activation of monocytes and cytokines. The severity of this response is due to the hydroxyl groups (OH) within the cellulose molecule. The replacement of these groups with acetate has resulted in modified cellulose triacetate membranes which reduce these responses and make them similar to synthetic dialysers. To determine the membrane's acute biocompatibility, complement activation and leukocytes were measured at 30min, and we found reductions similar to the published with synthetic dialysers. This led us to consider this membrane to be biocompatible and the concept that assumes that cellulose as bioincompatible was dismissed.15

There is a great heterogeneity of circulating monocytes in humans. Circulating monocytes which can be summarised as classical and non-classical according to whether they express greater or lower levels of CD14 and CD16 markers. Classical monocytes are most of the monocytes seen in healthy individuals. Non-classical monocyte populations tend are observed in certain pathological situations, such as cardiovascular disease16 or inflammatory processes.17 Several studies show a reduction in the percentage of non-classical monocytes and of markers of endothelial damage with OLHDF compared to conventional high-flux haemodialysis.18 Elevated levels of proinflammatory cytokines have been associated with greater mortality. Our results reveal that these membranes do not induce changes in these inflammatory markers after one month of treatment, as we have not found significant changes in the percentage of non-classical monocytes or in ILs levels. We therefore believe that this substituted cellulose membrane does not induce inflammatory changes different from those of normal synthetic membranes.

Our study has some limitations, although the number of patients is small, but we believe that the number of sessions performed is high and, therefore, the outcome is representative. The objective was not to compare this membrane with other membranes, but instead to describe the membrane it self. Therefore, the design is suitable. Although the number of sessions performed with some monitors, for example the Nikkiso monitor, was lower, we believe that the results are quite conclusive in terms of good adaptation of the filter to all systems. It is yet to be established if there is loss of albumin, which means that there is a question to be answered, however, we do not have the necessary resources to analyse this. We have found one report in which the loss of Albumin is quantified, and it is less than 2g per session.19

The new ATA® membrane obtains adequate Kt and Vconv values, without technical problems. We did not find the classic problems attributed to this type of membrane in the biocompatibility parameters studied, making it possible to perform OLHDF, and finding it to be most useful in patients who are allergic to synthetic membranes.

The study was funded by Nipro, with a financial contribution of €5000. This was allocated entirely to laboratory expenditure. None of the investigators received payment for the study.

The authors declare potential conflicts of interest.

Dr P. de Sequera, Dr M. Albalate and Dr R. Pérez García have participated as speakers at meetings organised by Gambro and Fresenius, and Dr G. Barril has participated at meetings organised by Nikkiso. They charged fees for these talks.