Introduction

Acute kidney injury (AKI) requiring renal replacement therapy (RRT) occurs in approximately 15% of ICU patients [1]. Intradialytic hypotension (IH) is a common adverse complication following RRT in critically ill patients regardless RRT modality. IH is associated with higher in-hospital mortality [2]. Intradialytic hypotension leads to renal perfusion impairment with further deterioration of renal function recovery in patients with AKI [3]. Analysis of patients with end-stage kidney disease receiving intermittent RRT suggests that some methods of IH correction (for example, cold dialysate with high concentration of sodium) can reduce the risk of renal failure. At the same time, modern systematic reviews devoted to correction of IH in critically ill patients with AKI do not demonstrate obvious advantages of a particular method [4]. In this regard, searching for effective methods of correction and prevention of IH and its consequences is still important. Biological incompatibility of the dialyzer membrane can provoke vasodilation (including anaphylaxis). It is one of the factors predisposing to IH. Biological incompatibility is followed by enhanced activation of leukocytes and complement in extracorporeal circuit, primarily on the dialyzer membrane, where blood comes into contact with non-physiological biomaterial [4, 5]. Biocompatibility is the most important feature of synthetic membranes compared to cellulose membranes. Synthetic polymer membranes (polymethyl methacrylate (PMMA), polyethersulfone, polysulfone (PS) or polyacrylonitrile (PAN)) are considered highly biocompatible and widely used in everyday clinical practice. These membranes minimize complement activation and blood inflammatory response in contrast to unmodified cellulose membranes used in the past [6]. For example, Furuta et al. [7] analyzed elderly patients with comorbidities and demonstrated that PAN membranes (AN69) for intermittent hemodialysis lead to a significantly lower incidence of IH and more effective clearance of inflammatory cytokines compared to PS-based dialyzers. Synthetic dialyzers are characterized by high ultrafiltration coefficient (> 25 ml / h / mm Hg / m²), high diffusion and convective performance, permeability for medium molecular weight metabolites. These dialyzers (for example, PMMA and PAN membranes) are able to adsorption of cytokines and endotoxin on their surface that is preferable in critically ill patients [8]. Adsorption characteristics largely determine biocompatibility. Therefore, the choice of membrane may also affect intradialytic hemodynamic stability that affects clinical results and achievement of the prescribed characteristics. To date, intermittent techniques in critically ill patients with AKI have not been sufficiently studied because prolonged renal replacement therapy (CRRT) is considered the optimal modality in critically ill patients with unstable hemodynamics [9]. At the same time, it is known that intermittent RRT is more cost-effective in short-term period compared to prolonged procedures requiring significant economic and organizational costs associated with the need for additional staff training and consumables (especially sterile substitute/dialysate) [10]. The most optimal RRT technique with minimum number of disadvantages in patients with AKI is unclear, and ongoing research in the future should bring some clarity.

The purpose of the study was to evaluate the effectiveness of various models of hemodiafilters in HDF-online regarding hemodynamic tolerance of the procedure and achievement of necessary clinical results in cardiac surgery patients with acute kidney injury and unstable hemodynamics.

Material and methods

The local ethics committee approved this prospective randomized study of 60 consecutive ICU patients with organ dysfunction, including AKI (KDIGO criteria [11]), after on-pump cardiac surgery and indications for RRT. All patients were randomized into 2 groups:

— group 1 (n=30) — HDF-online with isoosmolar dialysate (Na 140 mmol/l, t 37°C) and polysulfone FX800 dialyzer (Fresenius Medical Care, Germany);

— group 2 (n=30) — HDF-online with isoosmolar dialysate (Na 140 mmol/l, t 37°C) and PMMA BK-2.1U dialyzer (Toray Medical Co. Ltd., Tokyo, Japan).

All patients were ventilated and received vasopressor therapy with norepinephrine according to clinical indications. Clinical and anthropometric data were similar in both groups (Table 1).

Table 1. Clinical and anthropometric data in both groups

Note. * — data are presented as M±SD; NYHA — New York Heart Association; CPB — cardiopulmonary bypass; EuroSCORE — European System for Cardiac Operative Risk Evaluation; SOFA — Sequential Organ Failure Assessment.

Inclusion criterion was AKI KDIGO stage 2 (urine output <0.5 ml/kg/h for ≥12 hours or 2-fold increase in serum creatinine, baseline creatinine was considered the value at admission after surgery). Exclusion criteria: patients without postoperative AKI, inotropic support with two drugs and cardiac output < 2.5 l/min/m2, anuria requiring monitoring of volemic status.

HDF-online was performed using Fresenius 5008 dialysis machine (Fresenius Medical Care, Germany). Anticoagulation was ensured by intravenous administration of heparin under control of activated clotting time (ACT Plus device, Medtronic, USA). Target ACT was 180-200 sec. All patients underwent body weight control. Blood flow was 200 ml/min, dialysis flow — 500 ml/min. Substitute volume was automatically calculated in post-dilution mode (mean 68 ml/min or 49 ml/kg/h). Kt/V was 1.0-1.2 in hardware monitoring of the delivered dose. We applied acid concentrate for hemodialysis AC-F 313/1 and dry bicarbonate concentrate BIBAG 900g (Fresenius Medical Care, Germany). Ultrafiltration volume was determined individually depending on fluid balance and hemodynamic parameters (mean 30 ml/kg per a session). We performed invasive hemodynamic monitoring in all patients (systolic, diastolic and mean blood pressure, cardiac index, global end-diastolic volume index, extravascular lung water index) using the PiCCOplus monitor (Pulsion Medical Systems, Germany). For invasive hemodynamic monitoring by transpulmonary thermodilution, we catheterized femoral artery using a thermodilution catheter (5F, PV1520L20 PULSIOCATH, Pulsion Medical Systems, Germany). Thermodilution was achieved by 3 injections of 0.9% sodium chloride solution 15 ml cooled to 4 °C through a central venous catheter 8F. The last one was also used for administration of drugs and infusions. Indexed parameters were analyzed within the periods of hemodynamic stability before and immediately after HDF-online. Intradialytic systolic blood pressure < 90 mm Hg was regarded as IH and required correction (higher dosage of norepinephrine, bolus or decrease of ultrafiltration rate).

Statistical analysis

Normal distribution was tested using the Shapiro-Wilk test. Quantitative variables for control points "before RRT / after RRT" was analyzed using the Wilcoxon test for dependent samples. Medians were compared using the Friedman test adjusted for Holm multiple comparisons (pairwise post hoc comparisons were made using the Nemenyi test). Pairwise comparisons were made using the Fisher's exact test adjusted for Holm multiple comparisons. Differences were considered significant at p-value < 0.05. Statistical analysis was carried out using the R Foundation for Statistical Computing (Vienna, Austria, version 3.2).

Results

Parameters of hemo- and hydrodynamic status before and after RRT are presented in Table 2. We observed positive changes of volemic status in both groups (significant decrease of global end-diastolic volume index and extravascular lung water index with more obvious changes in the PMMA group). Changes of cardiac output were multidirectional. Indeed, there was a significant intradialytic decrease of cardiac output in the 1^st group and increase in the 2^nd group.

Table 2. Parameters of hemohydrodynamic status

Note. Data are presented as median [lower quartile; upper quartile]; between-group comparison using the Wilcoxon test; CI — cardiac index; GEDVi — global end-diastolic volume index; ELWI — extravascular lung water index; RRT — renal replacement therapy.

There were significant between-group differences in systolic blood pressure and norepinephrine dosages at baseline and after each hour of HDF-online (Fig. 1, 2).

Fig. 1. Dynamics of systolic blood pressure in the studied groups.

Fig. 2. Dynamics of the dose of norepinephrine in the study groups.

In the 1^st group, a more significant decrease of systolic blood pressure was associated with HDF-online and required higher doses of norepinephrine. In the 2^nd group, there was a stable hemodynamic profile and no need for vasopressor support by the end of the procedure.

Analysis of the incidence of IH also revealed significant between-group differences (Fig. 3).

Fig. 3. Incidence of intradialytic hypotension in both groups.

The 1^st group was characterized by significantly higher incidence of IH (systolic blood pressure < 90 mm Hg) compared to the 2^nd group that did not allow achieving the necessary clinical prescriptions (Table 3).

Table 3. Clinical characteristics of the procedure

Discussion

There is currently no consensus regarding the exact definition of intradialytic hypotension. Nevertheless, intradialytic systolic blood pressure decrease < 90 mm Hg is associated with high mortality, especially if baseline blood pressure was over 160 mm Hg [12]. Frequent episodes of IH deteriorate recovery of renal function in patients who survived AKI that will require continuation of RRT [13]. IH may be due to low arteriolar tone and peripheral vasodilation and reduced left ventricular contractility. Hemodialysis per se can deteriorate vascular tone and left ventricular contractility [14]. Reduced vascular tone may be caused by induction of cytokine release, bioincompatibility of dialytic membrane, use of acetate as a dialysate buffer, enhanced synthesis of nitric oxide or insufficient release of endogenous vasopressin during ultrafiltration [15]. In this regard, there are various strategies for prevention and treatment of IH based on the impact on one or more of the above-described factors [15]. Biocompatibility of dialytic membranes can be defined as the sum of specific interactions between blood and dialytic membranes, i.e. activation of complement, platelets, monocytes and neutrophils during RRT procedure [16]. The results of most clinical studies demonstrate the effectiveness of biocompatible dialytic membranes regarding improvement of survival and recovery of kidney function after AKI. Modern synthetic membranes (PS, PMMA or PAN) are considered to be biocompatible in contrast to cuprophane-based membranes used in the past, since they activate complement to a lesser extent [17]. Characteristics of transmembrane transport and biocompatibility are two important considerations when choosing a dialysis membrane. This significantly affects clinical outcomes in patients with AKI including the incidence of acute hemodynamic instability and anaphylactic reactions. Meanwhile, modern synthetic membranes also differ in their biocompatibility characteristics that can significantly affect clinical results [18]. However, beneficial effects of each membrane material have been evaluated in several small randomized clinical trials. As a rule, the authors compared regenerated cellulose with synthetic polymeric membranes. These results are not directly related to comparison of modern synthetic polymeric membranes. Nevertheless, the authors found no significant differences between synthetic polymeric membranes and regenerated cellulose membranes which are attributed to poor biocompatibility [19]. At the same time, PMMA-based membranes with an adsorption effect are considered to be highly effective in hypercytokinemic critical conditions and may be most useful, for example, in septic patients [19]. Patients with multiple organ failure after cardiac surgery are also characterized by significant increase of pro- and anti-inflammatory cytokines and their imbalance due to predominant pro-inflammatory cytokines [20]. Considering the above-mentioned concepts and own experience, we initiated our study in the belief that the choice of membrane could positively influence the efficacy of RRT and clinical outcomes in patients with AKI. All patients had unstable baseline hemodynamics and required vasopressor support with norepinephrine. Conditions of HDF-online were the same in all patients, and dialytic membranes differed only. In the 1^st group, there was a higher incidence (14 (47%) patients) of IH (systolic BP < 90 mm Hg) that required higher doses of norepinephrine, infusion bolus 300 ml and less rate of ultrafiltration. Two (6.67%) patients required discontinuation of the procedure. Heart rhythm disturbances (AF) developed in 9 patients (30%). Target ultrafiltration was achieved in 16 (53.3%) patients, the prescribed dose of dialysis (Kt / V) — in 28 (93.3%) patients. In the 2^nd group (PMMA), IH occurred only in one patient (3%). There were no cases of discontinuation of the procedure and rhythm disturbances. Target ultrafiltration and the prescribed dose of dialysis were achieved in all patients.

Considering ELWI data, we can talk about baseline interstitial fluid overload in both groups. Between-group analysis revealed significant decrease of ELWI, increment of cardiac output, normalization of GEDVi and no need for vasopressor support with norepinephrine by the end of HDF-online procedure in the PMMA group. In the PS group, there was a slight decrease of cardiac output, minimal reduction of ELWI, normalization of GEDVi with persistent need for norepinephrine infusion. In general, our results are consistent with the theory that characteristics of dialytic membranes can influence intradialytic hemodynamics. This phenomenon appears to be related to physicochemical and structural properties of these membranes. Obviously, further researches are required.

Conclusion

HDF-online with polymethyl methacrylate-based dialyzer was characterized by a minimal negative hemodynamic impact in critically ill patients with acute kidney injury compared to polysulfone dialyzers. This ensured the necessary clinical effects of renal replacement therapy (RRT). This type of RRT can be recommended in severe ICU patients.

The authors declare no conflicts of interest.