Online hemodiafiltration (HDF) versus high-flux hemodialysis (hf-HD): A review

Chronic kidney disease affects a vast part of the world population and as a consequence gradually declines renal function. Patients enter the end stage of kidney disease as the disease progresses, and the use of renal replacement therapy such as hemodialysis then becomes necessary. With the advance of technology, new modalities of dialyzers have been made available in the market with the objective of making the hemodialysis process more efficient and of increasing its biocompatibility. This review aims to discuss different hemodialysis techniques, focusing on online hemodiafiltration and high-flux hemodialysis, in terms of molecular clearance, biocompatibility, cardiovascular stability, survival, safety, and costs. In comparison to conventional hemodialysis (low-flux), online hemodiafiltration and high-flux hemodialysis present a greater capacity to filtrate medium molecular weight molecules, presenting greater biocompatibility and maintaining cardiovascular stability during dialysis sessions, and constituting factors which can justify better outcomes of patients submitted to these modalities. However, studies differ on the real superiority of online hemodiafiltration when compared to high-flux hemodialysis, which highlights the need for further discussion on the subject.


Introduction
The main function of the kidney is to eliminate toxins generated by the metabolic processes of the body (Clark et al., 2019). Patients with chronic kidney disease (CKD) or who are in end-stage renal disease (ESRD) have this function affected, which leads to accumulation of the aforementioned toxins in the circulatory system, a condition called "uremia" (Clark & Gao, 2002;Franco et al., 2019). An accumulation of these so-called uremic toxins (UTs) in the body generates several deleterious effects in the organism, including malnutrition, anemia, dermal atrophies, dyslipidemia, metabolic acidosis, inflammation, pericarditis, hypertension, heart failure, coagulation disorders, and cardiovascular disease (CVD). CVDs are the leading cause of death in these patients (Go et al., 2004;Vanholder et al., 2005, Vanholder et al., 2008. Patients in ESRD require renal replacement therapy (RRT) such as dialysis in order to prevent symptoms and to sustain their life . Despite technological and management advances to improve the quality of life (QoL) of patients in ESRD, it is notable that there is a high incidence of adverse complications such as hospitalization and death in this population, where the mortality rates of patients in ESRD are higher than patients with diabetes, CVD or cancer Saran et al., 2017). However, new hemodialysis techniques are being developed, improved and studied in order to improve this scenario. Conventional low-flux hemodialysis (HD) is not highly effective, and patient morbidity and mortality rates are still high (Locatelli et al., 2015). As an alternative, recent controlled trials demonstrate that dialysis techniques which make use of high-flux membranes or that associate diffusion with convection show superiority over conventional hemodialysis in clinical endpoints (Canaud et al., 2020).
Thus, this review aims to discuss the main characteristics and conclusions of the impact of high-flux hemodialysis (hf-HD) and online hemodiafiltration (HDF) techniques on molecular clearance, biocompatibility, cardiovascular stability, survival, safety and cost.

Methodology
The present study is a systematic exploratory review (Reyes, 2020) of scientific research and literature reviews conducted between 1999 and 2020, with the aim of discussing different hemodialysis techniques, focusing on online hemodiafiltration and high flux hemodialysis, in terms of molecular clearance, biocompatibility, cardiovascular stability, survival, safety and costs. MEDLINE/PubMed (U.S.A. National Library of Medicine), Web of Science: Science Citation Index (Clarivate Analysis), SciELO and Google Scholar databases were used to search for articles using the following descriptors: Hemodialysis, Hemodiafiltration, High Flux Hemodialysis, Uremic Toxins, Cardiovascular Stability, Filtration Membranes and Biocompatibility. Articles published before the year 1999 and those without the descriptors in the title and/or abstract were excluded, the remaining clinical and review studies were included in the review process ( Figure 1).

Hemodialysis Modalities
A suitable method for replacing renal function is HD, which is the most prevalent RRT technique worldwide. New HD modalities have emerged with the advancement in technology, aiming to improve the performance of this technique, as well as to increase the quality of life of patients and reduce the risks associated with the chronic use of RRT . Although any RRT type is available in most countries, it is limited to specific subgroups of the population and frequently the most vulnerable populations have no access to treatments nor to diagnostics.

High-flux hemodialysis (hf-HD)
HD is used to artificially remove circulating toxins by an external filter which contains a semipermeable membrane (Donadio et al., 2017;Vadakedath & Kandi, 2017) with a thickness of about 20 to 45 μm, length of 160 to 250 mm (Sakai, 2000), and pore size ranging from 1 to 14 nm. This dialysis modality (low-flux) mainly targets the filtration of low molecular weight UT (<500 Daltons), which limits its efficiency (Locatelli et al., 2010;Suchy-Dicey et al., 2016) since it makes use of diffusion (Ledebo, 1999;Ledebo & Blankestijn, 2010). The diffusive transport process requires the dialysis fluid, which flows through the counter-current from the dialyzer to blood ( Figure 2) (Ledebo, 1999;Ledebo & Blankestijn, 2010;Thomas & Jaber, 2009). The difference in solute concentration between the blood and the dialysate serves as a driving force for diffusive transport, and this force differs for each solute, the diffusion rate is defined by the molecular weight of the solute and its resistance to flow, this resistance is due to the hemodialysis membrane and its physical characteristics. The diffusion rate is inversely proportional to the cubic root of the molecular weight, which favors the clearance of low molecular weight molecules (Ingrid Ledebo & Blankestijn, 2010;Ledebo, 1999).
Alternatively, hf-HD aims to improve the clearance capacity using membranes with larger pores, high ultrafiltration and convective therapy, thus also removing medium molecular weight molecules (>500 Daltons and <12.000 Daltons). The convection process resembles the physiological process in the human kidney, whereby plasma is filtered through the glomerulus, which is selectively permeable, and glomerular filtration depends on a relatively low hydrostatic pressure, convection and solvent drag (Thomas & Jaber, 2009). The blood is filtered through a highly permeable biosynthetic membrane, promoting the removal of low and medium molecular weight molecules, which has a higher filtration capacity of low molecular weight UT (Meyer et al., 2005;Pecoits-Filho et al., 2019;Thomas & Jaber, 2009). In this modality, the total ultrafiltration can exceed the delimited weight loss and the compensation of this loss occurs within the dialyzer by retrofiltration. Convective transport is considerable due to the permeability of the high flux membranes, but cannot be controlled, which is the main difference when compared to online hemodiafiltration (HDF) (Ledebo & Blankestijn, 2010;Petrie et al., 2008). The convective transport volume must exceed 50 mL/min to have a significant impact on the clearance of medium molecular weight molecules, which does not normally occur in hf-HD (Ledebo, 1999;Lornoy et al., 2000). In addition, the biocompatibility of high-flux membranes reduces chronic inflammation and oxidative stress, which is strongly associated with malnutrition and atherosclerosis (Pieroni et al., 2015). High-flux dialysis seems to be an efficient alternative compared to other dialysis methods which make use of diffusion and convection, and a satisfactory compromise between efficiency and practicability (Donadio et al., 2017). However, there is no agreement on the equivalence of the techniques (i.e. studies demonstrate the superiority of hemodiafiltration technique when compared to hf-HD (Abad et al., 2016;Locatelli et al., 2018;Peters et al., 2016). Thus, further studies and clarifications on both techniques in terms of molecule clearance and interference in quality of life are necessary.

Online hemodiafiltration (HDF)
Previously called online haemofiltration was developed with the goal of improving filtration capacity of HD. This modality makes use of diffusion associated with convection and promotes the removal of medium molecular weight molecules, with this being the main advantage when compared to the conventional modality Thomas & Jaber, 2009). Weight loss control in this modality is done by infusing a replacement fluid, which has a similar composition to plasma water, and must be sterile and non-pyrogenic. This infusion can be administered before (pre-dilution) or after (post-dilution) the dialysis filter ( Figure 3) (Ledebo, 1999;Ledebo & Blankestijn, 2010;Petrie et al., 2008). can be administered before (pre-dilution mode) or after (post-dilution mode) the dialysis filter. This fluid has a composition similar to plasma water, is sterile and non-pyrogenic.

Source: Authors.
Online hemodiafiltration was first described in the 1970s, however it remains rarely used (Chuasuwan et al., 2020).
Several studies have compared HDF to other dialysis methods, however, Donadio et al. (2017) concluded that HDF is a more efficient treatment than low flux hemodialysis, but it is not accessible to all patients. According to Schiffl (2019), only 7% of patients worldwide are treated with HDF, while less than 1% of patients have access to this dialysis technique in the United States, as well as in Latin America.

The Clearance of Molecules
Uremic toxin clearance is directly related to kidney functions. The main renal clearance mechanisms are glomerular filtration and tubular secretion, which is characterized by transporter proteins that mediate the influx of uremic toxins through the basolateral membrane and subsequent efflux through the luminal membrane of renal tubule cells, in addition to being responsible for the reabsorption of these solutes (Suchy-Dicey et al., 2016;Wang & Kestenbaum, 2018). However, dialysis techniques in ESRD patients become essential instruments for the removal (at least partially) of uremic toxins from the body (van Gelder et al., 2020;Yamamoto et al., 2016). These compounds are classically divided according to physicochemical properties into three groups: small water-soluble, middle, and protein-bound molecules (Vanholder & Glorieux, 2003).

Small water-soluble molecules
Small water-soluble molecules are characterized by a molecular size less than 500 Da (Vanholder & Glorieux, 2003).
Renal clearance of TMAO is due to glomerular filtration and tubular secretion, in which it is a substrate of OCT1, OCT2, OAT3, and multiple efflux transporters (Gessner et al., 2018;Miyake et al., 2017;Teft et al., 2017;Wu et al., 2017). Small water-soluble molecule removal can occur in dialysis therapies via diffusion in semipermeable membranes. Hai et al. (2015) showed that the reduction rate of TMAO, urea and creatinine after a HD session could reach 86%, 77% and 71%, respectively.
In addition, Kim et al. (2019) reported similar removal percentages of small uremic toxins, such as urea, creatinine, uric acid, and phosphate, in the HDF modalities, high-flux dialyzers in HD (hf-HD), and medium cut-off (MCO) dialyzer in hemodialysis (MCO-HD).
More recently, MCO dialyzers were developed to obtain optimized permeability, with larger pores which allow the passage of middle molecules (up to about 50 kDa) and minimal albumin loss (Boschetti-De-Fierro et al., 2015;Wolley et al., 2018). Some clinical studies have shown that MCO-HD and HDF is able to more effectively remove middle molecules and a wider range of middle-molecule uremic toxins compared to HF-HD and HD (Belmouaz et al., 2020;Kirsch et al., 2017;Locatelli et al., 2015;Weiner et al., 2020). Belmouaz et al. (2020) demonstrated that MCO-HD had reduction rates of β2-microglobulin and FGF-23 of 73% and 41%, respectively, which is significantly higher than hf-HD. However, the use of MCO membranes in large populations and their possible clinical effects remain to be elucidated.
However, the removal of these compounds by dialysis therapies is limited due to protein binding capacity. Itoh et al. (2012) demonstrated that the reduction rate of indoxyl sulfate and p-cresyl sulfate after a HD session was only 31.8% and 29.1%, respectively. Some studies have shown that molecules (i.e. ibuprofen) are able to displace uremic toxins by competing for the same binding site on albumin, which leads to an increase in the free fraction of these uremic toxins and improve clearance by dialysis therapies (Madero et al., 2019;Shi et al., 2019;Tao et al., 2016). Madero et al. (2019) reported that the ibuprofen infusion into the bloodstream prefilter increased the indoxyl sulfate and p-cresyl sulfate concentration in the outflow dialysate Research, Society and Development, v. 11, n. 4, e17111427237, 2022 (CC BY 4.

Biocompatibility of Hemodialysis Membranes
Hemodialysis is a procedure performed in extracorporeal circulation, and the lack of biocompatibility may induce significant clinical and biochemical responses. Due to blood contacting with the components of the hemodialysis circuit, such as dialysis membranes, drains and intravenous fluids administered during the procedure, activation of inflammatory and immune pathways may occur, leading to dialysis-induced oxidative stress and membrane-induced inflammation, which are related to cardiovascular problems (Chang et al., 2014;Gomółka et al., 2020;Ojeda et al., 2020;Claudio Ronco et al., 2018;Claudio Ronco & Clark, 2018). This activation occurs via the alternative pathway of the complementary system, leading to anaphylatoxins being released into the bloodstream, activation of the coagulation cascade and of different cell types (monocytes, neutrophils, basophils, and platelets), and elevating plasma pro-inflammatory cytokine levels such as interleukin-1β (IL-1), IL-6, IL-10, tumor necrosis factor-α (TNF-α) and stimulating the production of proteolytic enzymes by granulocytes Due to the exposure time of the blood to the dialysis membrane surface, the biocompatibility of this HD component is crucial. Membranes have the function of removing water and solutes through various mass separation processes such as diffusion, convection and adsorption (Ojeda et al., 2020;Claudio Ronco & Clark, 2018). These membranes can be traditionally classified according to their composition and permeability, being made of cellulose (modified or unmodified) or synthetic.
Unmodified cellulose membranes were the most common in the past, but they were eradicated from the market, and the same has been observed for modified cellulose membranes which have good performance and functionality when compared to synthetic ones; however, their biocompatibility and permeability are limited, preventing their application in convective therapies (Clark et al., 1999;Gomółka et al., 2020;Kalantar-zadeh et al., 2018;Ojeda et al., 2020;Claudio Ronco & Clark, 2018).
These limitations have been surpassed with the advance of biomaterials and with the improvement in the production of fibers that have resulted in new membranes with specific characteristics and refined properties, which have begun to present greater biocompatibility and permeability, in addition to expanding their categorization. Factors such as permeability index, hydrophilic index, adsorption capacity, and electric potential are now considered to define the membrane category Ronco & Clark, 2018). Studies show that the use of synthetic high-flux membranes have greater biocompatibility by reducing chronic inflammation and oxidative stress, as in the case of membranes used in hf-HD and HDF (Clark et al., 1999;Gomółka et al., 2020;Ojeda et al., 2020;Pieroni et al., 2015;Tomo, 2016).

Morena et al. (2019) compared hf-HD and HDF in relation to their biocompatibility and noted no difference between
the two methods, with no change in inflammatory parameters (IL-6 and TNF-α) nor in albumin concentration observed in samples from 32 patients submitted to the two dialysis methods and using four different types of hemodialysis membranes. Gomółka et al. (2020) also reaffirmed the adequate biocompatibility of these two dialysis methods by analyzing samples from

Cardiovascular Stability
The most frequent clinical problem in patients during HD is cardiovascular (CV) instability, which occurs due to a decreasing blood volume during filtration, leading to vasoconstriction and reduced perfusion in some body regions, triggering systemic circulatory stress (Locatelli et al., 2018;Penny et al., 2018;Ronco et al., 2000). A patient may experience stunting of the myocardium and of other vital organs or precipitated ischemia as a consequence of this instability (Locatelli et al., 2018;Ok et al., 2013;Penny et al., 2018). Studies point out that it is necessary to accurately monitor possible causes of cardiovascular injuries during HD in order to prevent these cardiovascular events, and provide effective preventive intervention for such injuries (Penny et al., 2018).
Despite studies pointing towards HDF and hf-HD as alternatives to HD to minimize the occurrences of these cardiovascular events, the clinical benefits of HDF compared to hf-HD continue to be debated. Articles highlight both the similarity of the techniques and the superiority of HDF over hf-HD (Table 1) (La Milia et al., 2019;Locatelli et al., 2015Locatelli et al., , 2018Shroff et al., 2018). La Milia et al. (2019) stated that the removal of Na + balance and plasma tonicity did not differ between hf-HD and HDF and this was not a factor which could explain the superiority of HDF. Studies have evidenced that HDF can improve cardiovascular stability when compared to hf-HD (Locatelli et al., 2018) in patients receiving higher convection volumes, however this limits the reach of this technique, since patients who have suboptimal vascular access or time limitations associated with dialysis may not reach these convection volumes (Grooteman et al., 2012;Ok et al., 2013;J. R. Smith et al., 2017). Compared to HF-HD, MCO-HD induces a higher myoglobin reduction ratio, but also reduces the pre-dialysis levels of other medium molecules, including beta2-microglobulin, and kappa and lambda free light chain. Thus, MCO-HD appears as a strategy in the removal of medium-molecular-weight toxins, and can be considered an alternative to HDF. (van Gelder et al., 2020) Netherlands

2004-2009
Randomized multicentercontrolled trial 80 HDF Clearance of PBUTs, mortality and cardiovascular events HDF did not consistently decrease plasma PBUT concentrations in patients and no relationship was found between PBUTs and cardiovascular endpoints and mortality. (Kim et al., 2019)

Republic of Korea 2020
Observational prospective study 6 MCO-HD vs hf-HD vs pre-HDF Clearance of larger middle molecules MCO-HD showed significantly greater clearance of large medium molecules and achieved better clearance of and kappa and lambda free light chain than hf-HD and HDF, without the need for large convection volumes or high blood flow rates. MCO-HD presents an advantage for older adult patients with vascular access deficiencies and HD patients without access to HDF. (Kirsch et al., 2017) Austria and Germany 2015 Prospective, open-label, 4arm, randomized, active control, crossover pilot study 20 MCO-HD vs hf-HD Clearance of larger middle molecules MCO HD removed a wide range of molecules from the medium more effectively than high-flux HD and even exceeds the performance of high-volume HDF for large solutes, particularly kappa free light chain. (Donadio et al., 2017) Italy 2017 Randomized cross-over study 30 HD vs hf-HD vs HDF Safety, efficiency, and removal mechanisms of toxins The clearance of small toxins was similar between hf-HD and HD. β2-microglobulin was only removed with hf-HD, which had excellent tolerability. The efficiency of hf-HD was similar to HDF. Proteomic analysis demonstrated that only high-flow membranes remove and adsorb small proteins. hf-HD may be an efficient alternative to HDF. (Locatelli et al., 2018)  hf-HD vs HDF Self-reported sleep duration The two dialysis modalities had no effect on self-reported sleep duration.

Survival
Renal function impacts the QoL of patients in ERSD and, like the use of RRT, interferes with morbidity and mortality (Perl et al., 2017;Peters et al., 2016). Studies based on cross-sectional and validated measures of health-related QoL among patients undergoing HD show a dramatic reduction in QoL when compared to the general population and to patients with other chronic diseases. Therefore, care of HD patients includes providing adequate nutrition as well as sufficient dialysis, maintaining vascular access, minimizing hospitalizations, and most importantly improving or maintaining their quality of life (Table 1) (Abdelsalam et al., 2020).
The high morbidity and mortality rates in patients undergoing HD have been associated with inadequate rates of the removal of medium molecular weight molecules, which has influenced the implementation of new HD techniques such as hf-HD or HDF and of using new classes of filtration membranes (Morena et al., 2019;Okuno et al., 2009;Peters et al., 2016). Eknoyan et al. (2002) demonstrated that hf-HD significantly impacts survival in the long term, with a 32% reduction in mortality in patients who participated in the study. Other studies also evidence an interference of hf-HD in decreasing the incidence of complications, treatment duration, and in improving survival (Table 1)  Studies in the literature on the influence of HDF on patient survival diverge (Locatelli et al., 2018). Articles state that despite providing additional elimination of medium molecular weight toxins compared to HD, HDF is not associated with a lower mortality risk (Peters et al., 2016). However, other clinical studies in dialysis patients have shown benefits of this technique in relation to survival. Peters et al. (2016) noted that patients undergoing HDF had a considerably lower risk of allcause mortality than those receiving HD, while the risk of long-term CVD mortality in HDF patients was 31% lower than in patients receiving HD. Compared to hf-HD, there is no consensus on the superiority of HDF (Figure 4).  (Chuasuwan et al., 2020;Ok et al., 2013;Shroff et al., 2018;Den Hoedt et al., 2014, andVilla et al., 2011), HDF to HD ratios were obtained and then converted to HDF to hf-HD ratios based on reference (Ojeda et al., 2020), which compared all-cause mortalities among these 3 dialysis methods.

Safety and Cost
HD is a costly treatment that is essential for ESRD patients (Villa et al., 2011). Safety and costs aspects of implementing HD modalities are relevant and could have an impact on public and private health systems, especially considering the increase in the prevalence and incidence of CKD worldwide (Bikbov et al., 2020;Villa et al., 2011).
Regarding safety, the International Organization for Standardization (ISO) 23500-5:2019 and other regulatory guidelines specify the quality aspects, such as chemical and microbiologic quality standards, for dialysis fluids used in HD, HDF and related therapies (Kawasaki et al., 2009). Historically, failures in the quality control system in dialysis have led to contamination outbreaks, which has diminished over time due to best practices and regulatory standards (Coulliette & Arduino, 2013;Roth & Jarvis, 2000). HDF uses especially large convection volumes and the substitution fluid needs to be sterile and non-pyrogenic as it is infused into the patient, and is therefore important to ensure a safe and effective procedure (Tattersall & Ward, 2013;Ward et al., 2018). Clinical studies have supported the use of HDF, indicating quality and microbiological control (Bolasco et al., 2013;Penne et al., 2009;Vaslaki et al., 2000). Moreover, ultrapure dialysate reduced markers of inflammation and oxidative stress (Susantitaphong et al., 2013).
The cost-effectiveness of low-flux HD, high-flux HD, and HDF modalities is comparatively complex considering the quality-adjusted life year (QALY), various economic aspects, and methodologies used. Lévesque et al. (2015) demonstrated that high efficiency HDF is cost-effective compared to low-flux HD. Takura et al. (2013) also suggest that HDF is costeffective. Similarly, Ramponi et al. (2016) found that HDF is cost-effective compared to hf-HD, particularly among younger patients. However, Mazairac et al. (2013) reported that HDF cannot be considered a cost-effective modality when compared to HD, even in evaluating costs per QALY.

Final Considerations
Hemodialysis has proven to be a viable and important alternative in treating ESRD patients, even with the limitations involved in the initial modalities of renal function replacement. In this review, the definition of hemodialysis, as well as hf-HD and HDF, and their involvement in molecular clearance, cardiovascular stability, and survival were described. Information on biocompatibility of hemodialysis membranes, safety and cost were also discussed. Despite the existence of clinical studies, the superiority of HDF compared to hf-HD is not yet clear, and new clinical studies should be designed in order to clearly prove the benefits of HDF with respect to molecular clearance and mortality.