**2. HDF strategies that do not require exogenous substitution infusion**

Hemodiafiltration is an intermittent renal supportive therapy that involves the process of convection and diffusion. Total filtration volumes invariably exceed desired amounts and this dehydration must be corrected in real time. Despite various modifications of the HDF techniques based on infusion modes, the need for external replacement fluid infusion has not been eliminated. Accordingly, efforts continue to be made to eliminate exogenous sterile fluid infusion during HDF sessions. This is achieved by spontaneous fluid reinfusion at a rate that matches convection. Backfiltration and regenerated ultrafiltrate can be the methods of spontaneous fluid restoration.

### **2.1 Internal Filtration Enhanced HDF (internal HDF or iHDF)**

Internal filtration (IF) is defined as the total water flux across membranes within the closed blood and dialysate compartments of a dialyzer (Dellanna et al., 1996). Volume controlled high-flux HD is a representative modality to use the internal filtration phenomenon, and provides a straightforward means of achieving enhanced convection by augmenting internal filtration rates. The amount of internal filtration is directly regulated by pressure gradients through the hemodialyzer. A pressure drop is inevitable, as fluid flows through a cylindrical tube, and it is expressed by Poiseuille's equation:

Therefore, HDF is considered the gold standard for high-dose convective therapy, and has even been reported to reduce mortality risk as compared with high-flux HD (Canaud et al., 2006). HDF, which describes an intermittent renal supportive therapy of combined simultaneous diffusive and convective solute transport, is characterized by a large filtration volume that far exceeds the desired volume removal, and hence, external substitution is essential. In early HDF trials, a large number of sterile bags were used to supply substitution fluid, which was costly and complicated (Ledebo, 2007). However, technical advances made in the production of pyrogen-free ultrapure water allow sterile dialysate to be readily produced, which enables on-line based HDF to be used on a clinical basis. Several types of on-line HDF are clinically available that differ in terms of the ways in which external replacement fluid is administered, such as, by pre- or postdilution. Due to their unique benefits, mixed forms of pre- and post-infusion have also been devised, such as, mixed-dilution or mid-dilution HDF (Krieter & Canaud, 2008, Pedrini & De Cristofaro, 2003). However, the inevitable complexities associated with HDF machines and patient monitoring, and the requirement for the exogenous infusion of replacement fluid is still problematic. Therefore, various modifications of HDF strategies have been proposed to integrate HDF and HD modes, that is, to increase convective dose without the requirement for external infusion. These modifications can be classified into three developmental categories; (1) to increase the internal filtration rate by increasing pressure gradients along the hemodialyzer, (2) to use independent domains for forward filtration and backfiltration, or ultrafiltration and diffusion, and (3) to alternate forward

In this chapter, the trials on HDF strategies undertaken without exogenous substitution infusion will be discussed in terms of their technical aspects, *in vivo* and *in vitro* efficacies and applicabilities for clinical use. This is followed by an in-depth review on pulse push/pull hemodialysis (PPPHD), a recently introduced pulsatile technique that provides

Hemodiafiltration is an intermittent renal supportive therapy that involves the process of convection and diffusion. Total filtration volumes invariably exceed desired amounts and this dehydration must be corrected in real time. Despite various modifications of the HDF techniques based on infusion modes, the need for external replacement fluid infusion has not been eliminated. Accordingly, efforts continue to be made to eliminate exogenous sterile fluid infusion during HDF sessions. This is achieved by spontaneous fluid reinfusion at a rate that matches convection. Backfiltration and regenerated ultrafiltrate can be the methods

Internal filtration (IF) is defined as the total water flux across membranes within the closed blood and dialysate compartments of a dialyzer (Dellanna et al., 1996). Volume controlled high-flux HD is a representative modality to use the internal filtration phenomenon, and provides a straightforward means of achieving enhanced convection by augmenting internal filtration rates. The amount of internal filtration is directly regulated by pressure gradients through the hemodialyzer. A pressure drop is inevitable, as fluid flows through a cylindrical

**2. HDF strategies that do not require exogenous substitution infusion** 

**2.1 Internal Filtration Enhanced HDF (internal HDF or iHDF)** 

tube, and it is expressed by Poiseuille's equation:

and backward filtration procedures.

of spontaneous fluid restoration.

infusion-free HDF.

Fig. 1. Blood and Dialysate Pressure Gradient along Dialyzer Length. The sum of hydraulic and osmotic pressures is termed TMP, as TMP = ∆Pb- ∆Pd-∆π. Here, ΔPb represents the average value of arterial and venous blood pressure, ΔPd for average hydraulic dialysate pressures, and Δπ is oncotic pressures.

$$
\Delta P \propto \frac{\mu L}{d^4} Q \tag{1}
$$

Where, ΔP is the pressure drop, µ is the fluid viscosity, L and d are the length and diameter of the flow path, and Q the flow rates. Thus, blood and dialysate pressures drop along the dialyzers. However, because blood and dialysate flow in opposite directions, these pressure drops occur with opposing gradients, and in some region, hydraulic blood and dialysate pressures overlap (Fig. 1). In a normal countercurrent dialysis setup, the sum of hydraulic and osmotic pressures, termed transmembrane pressure (TMP), is positive in the proximal region of a hollow fiber dialyzer, and plasma moves to the dialysate compartment across the membranes (forward filtration). However, fluid movement occurs in the opposite direction in the distal region, because hydraulic blood pressures are below the sum of dialysate compartment pressure and osmotic pressure, and thus, backward filtration occurs and compensates for fluid loss in the proximal region.

#### **2.1.1 Factors influencing internal HDF**

Even though forward and backward filtration rates are highly dependent on membrane permeabilities and the degree of membrane fouling, they remain directly proportional to the positive and negative TMPs, respectively. As shown in Fig. 1, resulting TMP gradients can be readily increased by increasing blood and/or dialysate pressure drops (Fiore & Ronco, 2004, 2007). For blood, the pressure drop is proportional to blood viscosity and tube length in accord with Poiseuille's equation (Eq. 1), which shows that tube length increases pressure

Pulse Push/Pull Hemodialysis: Convective Renal Replacement Therapy 117

unique design of housing wall, involving addition of a short taper to the inner housing surface effectively prevented the dialysate from being channeled (Fujimura et al., 2004).

The beneficial effects of internal HDF, such as increased middle-size solute clearances, may be quantified by evaluating internal filtration rates, because the internal filtration level is directly related to delivered convective dose. The flux across a membrane (Q) at a local region of the hemodialyzer (x) can be expressed by membrane hydraulic permeability (Kuf

*Q x Kuf x TMP x Kuf x Pb Pd* \* ) \*(

radio-labeled molecules and the complexities of the procedures and equipment.

Another approach to determine internal filtration is offered by Doppler ultrasonography, which measures changes in blood velocity within dialyzers. In the absence of net-filtration, blood volume depletion in the proximal portion of a hemodialyzer leads to a reduction in blood flow velocity, and after the lowest point has been reached, the blood velocity gradually increases due to backfiltration. Thus, changes in blood velocity along a dialyzer provide information on blood volume changes and on amounts of forward and backward filtration. In one study, the internal filtration rate of a 250 mm dialyzer was found to be 37.7

Where, Pb, Pd and π represent the blood, dialysate and osmotic pressures. However, the flow dynamics inside the hemodialyzer are so complex that precise determinations of internal filtration rates are not available clinically. This is principally because Kuf across membranes is neither linearly related to the pressure gradient, nor constant at any position in the hemodialyzer (Ficheux et al., 2010). Kuf values are also substantially lowered by membrane fouling, which is remarkably affected by blood viscosity, coagulation, the abilities of membrane materials to bind plasma proteins, and treatment modalities. Hence, fluxes and permeabilities across membranes become parameters beyond the operator's control. Alternatively, a semi-empirical model based on clinical data has been developed to determine internal filtration rates. Using this model, internal filtration volumes and reinfusion rates were determined during internal HDF and post-dilution HDF modes, and revealed that differences between total convections (4.1 and 5.4 L/h for iHDF and HDF) well reflected differences between β2M clearance rates (123±11 and 149±26 ml/min for iHDF and HDF, respectively) (Lucchi et al., 2004). In a study conducted using *in vitro* scintigraphy method to verify this semi-empirical model, the model was found to show excellent accuracies of around 97% and a prediction error of only 3% (Fiore et al., 2006). In addition to the mathematical model, methods for performing indirect measurements of the internal filtration have also been proposed. Changes in non-permeable molecular concentrations occur in response to the water content of blood, and thus, the kinetics of water transport across membrane can be evaluated by measuring the cumulative concentration changes of non-permeable molecules (Ronco et al., 1992). Radiolabeled albumin (a non-permeable molecule) has been employed to determine the amounts of convection for hemodialyzers with reduced fiber diameters or an obstacle in dialysate stream (Ronco et al., 2000, 1998). A series of *in vitro* experiments proved that this scintigraphic method was accurate for measuring internal filtration rates, but despite its precision, its clinical application is not plausible due to the safety issue raised by the use of

(2)

**2.1.2 Internal filtration quantification** 

measured in ml/h/mmHg) and TMP (mmHg), as follows:

differential. Likewise, blood hematocrit and total protein levels also affect the pressure drop through viscosity.

The diameter of the flow path is another important factor. Poiseuille's equation shows that the pressure drop is inversely proportional to the 4th power of tube diameter, which means that membrane (a bundle of hollow fibers) lumen diameter is the predominant factor for governing blood pressure drops, and therefore, many investigations of internal HDF have focused on dialytic efficiencies using hemodialyzers with smaller membrane diameters. In early clinical studies, beta-2 microglobulin (β2M) removal was found to be significantly increased when membrane diameter was reduced. A 175 µm diameter dialyzer was found to enhance β2M clearances by factors of two and four, respectively, over 200 and 250 µm diameter hemodialyzers (Dellanna et al., 1996). Clearances of inulin and vitamin b12 were also significantly greater with 175 µm dialyzer than a 200 µm dialyzer, without changing the clearances of low molecular weight solutes (Ronco et al., 2000). In addition, a mathematical model showed that internal filtration rates increase rapidly with membrane diameter, and this theoretical result was also confirmed experimentally (Mineshima, 2004, Mineshima et al., 2000). Myoglobin clearance was increased by 34% when a membrane of diameter 150 µm, rather than 200 µm, with the same surface area was used at the same blood flow rates. These benefits in dialytic efficiency afforded by reducing membrane lumen diameter allow internal filtration enhanced hemodialyzers to be used clinically (Lucchi et al., 2004, Righetti et al., 2010).

However, the underlying risk of hemoconcentration due to high levels of forward filtration may not be negligible. Pressure-driven filtration causes large volume losses from blood and promptly increases blood viscosity, which deteriorates membrane sieving and hydraulic capabilities. Membrane-binding by blood components is a major cause of permeability reductions, and inevitably diminish membranes efficiencies, particularly in the forward filtration region. Nevertheless, membrane fouling, which tends to be more of an issue during the early stage of iHDF treatment, tends to have little effect on overall membrane transfer capacity during iHDF (Yamamoto et al., 2005).

Dialysate pressure is also regulated by increasing the flow resistance on the dialysate stream. Several techniques can be used to increase dialysate flow resistances, such as increasing membrane packing density, lengthening the hemodialyzer, or placing obstacles in the dialysate flow path. Obviously, dialyzer length effectively regulates dialysate pressure drops. In one study conducted to clarify the effect of dialyzer length on solute clearance, middle-to-large uremic molecules, such as β2M and alpha-1 microglobulin (α1M), were shown to be better cleared by a 250 mm dialyzer than a 195 mm dialyzer (Sato et al., 2003). Dialysate pressure drop can also be manipulated by modulating membrane packing density. The higher the packing density of membrane fibers, the greater the resistance to dialysate flow, because the effective cross-sectional area available for dialysate flow decreases. Analytical and experimental studies revealed myoglobin clearances using a hemodialyzer with 71.3% packing density were slightly higher than hemodialyzers with packing densities of 52% or 60.1% (Mineshima, 2004). However, high hemodialyzer packing densities cause substantial degrees of dialysate channeling and flow mismatch between blood and dialysate. This unmatched flow distribution leads to a loss of effective surface area and impairs the diffusion process (Gastaldon et al., 2003). Flow visualization studies in a dialyzer with a high packing density (75%) reconfirmed this disproportionate flow pattern of dialysate, as compared with standard packing density dialyzers (68%), and consequent reductions in urea clearance (Fujimura et al., 2004, Yamamoto et al., 2005). Nevertheless, a unique design of housing wall, involving addition of a short taper to the inner housing surface effectively prevented the dialysate from being channeled (Fujimura et al., 2004).

### **2.1.2 Internal filtration quantification**

116 Progress in Hemodialysis – From Emergent Biotechnology to Clinical Practice

differential. Likewise, blood hematocrit and total protein levels also affect the pressure drop

The diameter of the flow path is another important factor. Poiseuille's equation shows that the pressure drop is inversely proportional to the 4th power of tube diameter, which means that membrane (a bundle of hollow fibers) lumen diameter is the predominant factor for governing blood pressure drops, and therefore, many investigations of internal HDF have focused on dialytic efficiencies using hemodialyzers with smaller membrane diameters. In early clinical studies, beta-2 microglobulin (β2M) removal was found to be significantly increased when membrane diameter was reduced. A 175 µm diameter dialyzer was found to enhance β2M clearances by factors of two and four, respectively, over 200 and 250 µm diameter hemodialyzers (Dellanna et al., 1996). Clearances of inulin and vitamin b12 were also significantly greater with 175 µm dialyzer than a 200 µm dialyzer, without changing the clearances of low molecular weight solutes (Ronco et al., 2000). In addition, a mathematical model showed that internal filtration rates increase rapidly with membrane diameter, and this theoretical result was also confirmed experimentally (Mineshima, 2004, Mineshima et al., 2000). Myoglobin clearance was increased by 34% when a membrane of diameter 150 µm, rather than 200 µm, with the same surface area was used at the same blood flow rates. These benefits in dialytic efficiency afforded by reducing membrane lumen diameter allow internal filtration enhanced hemodialyzers to be used clinically (Lucchi et al., 2004, Righetti

However, the underlying risk of hemoconcentration due to high levels of forward filtration may not be negligible. Pressure-driven filtration causes large volume losses from blood and promptly increases blood viscosity, which deteriorates membrane sieving and hydraulic capabilities. Membrane-binding by blood components is a major cause of permeability reductions, and inevitably diminish membranes efficiencies, particularly in the forward filtration region. Nevertheless, membrane fouling, which tends to be more of an issue during the early stage of iHDF treatment, tends to have little effect on overall membrane

Dialysate pressure is also regulated by increasing the flow resistance on the dialysate stream. Several techniques can be used to increase dialysate flow resistances, such as increasing membrane packing density, lengthening the hemodialyzer, or placing obstacles in the dialysate flow path. Obviously, dialyzer length effectively regulates dialysate pressure drops. In one study conducted to clarify the effect of dialyzer length on solute clearance, middle-to-large uremic molecules, such as β2M and alpha-1 microglobulin (α1M), were shown to be better cleared by a 250 mm dialyzer than a 195 mm dialyzer (Sato et al., 2003). Dialysate pressure drop can also be manipulated by modulating membrane packing density. The higher the packing density of membrane fibers, the greater the resistance to dialysate flow, because the effective cross-sectional area available for dialysate flow decreases. Analytical and experimental studies revealed myoglobin clearances using a hemodialyzer with 71.3% packing density were slightly higher than hemodialyzers with packing densities of 52% or 60.1% (Mineshima, 2004). However, high hemodialyzer packing densities cause substantial degrees of dialysate channeling and flow mismatch between blood and dialysate. This unmatched flow distribution leads to a loss of effective surface area and impairs the diffusion process (Gastaldon et al., 2003). Flow visualization studies in a dialyzer with a high packing density (75%) reconfirmed this disproportionate flow pattern of dialysate, as compared with standard packing density dialyzers (68%), and consequent reductions in urea clearance (Fujimura et al., 2004, Yamamoto et al., 2005). Nevertheless, a

transfer capacity during iHDF (Yamamoto et al., 2005).

through viscosity.

et al., 2010).

The beneficial effects of internal HDF, such as increased middle-size solute clearances, may be quantified by evaluating internal filtration rates, because the internal filtration level is directly related to delivered convective dose. The flux across a membrane (Q) at a local region of the hemodialyzer (x) can be expressed by membrane hydraulic permeability (Kuf measured in ml/h/mmHg) and TMP (mmHg), as follows:

$$Q(\mathbf{x}) = \mathbf{K}\mu f(\mathbf{x})^\* \, TMP\{\mathbf{x}\} = \mathbf{K}\mu f\left(\mathbf{x}\right)^\* \left(\Delta P b - \Delta P d - \Delta \pi\right) \tag{2}$$

Where, Pb, Pd and π represent the blood, dialysate and osmotic pressures. However, the flow dynamics inside the hemodialyzer are so complex that precise determinations of internal filtration rates are not available clinically. This is principally because Kuf across membranes is neither linearly related to the pressure gradient, nor constant at any position in the hemodialyzer (Ficheux et al., 2010). Kuf values are also substantially lowered by membrane fouling, which is remarkably affected by blood viscosity, coagulation, the abilities of membrane materials to bind plasma proteins, and treatment modalities. Hence, fluxes and permeabilities across membranes become parameters beyond the operator's control. Alternatively, a semi-empirical model based on clinical data has been developed to determine internal filtration rates. Using this model, internal filtration volumes and reinfusion rates were determined during internal HDF and post-dilution HDF modes, and revealed that differences between total convections (4.1 and 5.4 L/h for iHDF and HDF) well reflected differences between β2M clearance rates (123±11 and 149±26 ml/min for iHDF and HDF, respectively) (Lucchi et al., 2004). In a study conducted using *in vitro* scintigraphy method to verify this semi-empirical model, the model was found to show excellent accuracies of around 97% and a prediction error of only 3% (Fiore et al., 2006).

In addition to the mathematical model, methods for performing indirect measurements of the internal filtration have also been proposed. Changes in non-permeable molecular concentrations occur in response to the water content of blood, and thus, the kinetics of water transport across membrane can be evaluated by measuring the cumulative concentration changes of non-permeable molecules (Ronco et al., 1992). Radiolabeled albumin (a non-permeable molecule) has been employed to determine the amounts of convection for hemodialyzers with reduced fiber diameters or an obstacle in dialysate stream (Ronco et al., 2000, 1998). A series of *in vitro* experiments proved that this scintigraphic method was accurate for measuring internal filtration rates, but despite its precision, its clinical application is not plausible due to the safety issue raised by the use of radio-labeled molecules and the complexities of the procedures and equipment.

Another approach to determine internal filtration is offered by Doppler ultrasonography, which measures changes in blood velocity within dialyzers. In the absence of net-filtration, blood volume depletion in the proximal portion of a hemodialyzer leads to a reduction in blood flow velocity, and after the lowest point has been reached, the blood velocity gradually increases due to backfiltration. Thus, changes in blood velocity along a dialyzer provide information on blood volume changes and on amounts of forward and backward filtration. In one study, the internal filtration rate of a 250 mm dialyzer was found to be 37.7

Pulse Push/Pull Hemodialysis: Convective Renal Replacement Therapy 119

Fig. 2. Schematic Pressure Profiles during Double HDF, when a flow restrictor is placed on

TMP regulation is also achieved by regulating dialysate pressure. Flow resistance applied to the dialysate tubing between the two dialyzers promptly increases dialysate pressures at the venous dialyzer because blood and dialysate flow in opposing directions. Hydraulic dialysate pressures exceed blood pressures, which leads to backfiltration in the venous dialyzer. However, dialysate pressures rapidly fall in the arterial dialyzer due to flow restriction, which causes ultrafiltration in the arterial dialyzer. In addition, the high blood and dialysate flow rates used are also associated with larger pressure gradients. Hence, ultrafiltration at the arterial dialyzer at levels of exceeding those required can be promptly compensated for by backfiltration at the venous dialyzer, and thus, exogenous replacement

The flow resistance placed on the dialysate stream was originally made from a gauge needle assembled with a bypass line in parallel. A clamp on the bypass line forced the dialysate into the gauge needle, and created flow resistance in dialysate stream. The flow resistance in this configuration is fixed, and the amounts of ultrafiltration and backfiltration cannot be adjustable externally. Hence, the means of creating resistance to dialysate flow was improved in the advanced version, termed convection-controlled double high-flux HDF, in which variable and controllable flow resistances were integrated (Pisitkun et al., 2004). Therefore, together with these features, this modality achieved unmatched depurative outcomes, as demonstrated by far higher uremic molecular clearances regardless of molecular size (Cheung et al., 1982, Shinzato et al., 1982, von Albertini et al., 1985). Furthermore, increased clearances allowed treatment times to be shortened (Miller et al.,

blood tube (upper) and on dialysate tube (below) of the two hemodialyzers.

infusion is not required for this method.

ml/min by Doppler ultrasonography, but only 11.1 ml/min for a standard 195 mm dialyzer (Sato et al., 2003). Doppler ultrasonography is straightforward, non-invasive, and easily used at bedside (Mineshima, 2011). However, the method is still incapable of measuring blood flow velocity precisely, particularly blood velocity deep within the membrane fiber bundle. In other words, this method is based on velocities measured in peripheral membranes, which are quite different from velocities within centrally located fibers, and as a result, deviations from true values are unavoidable.

Other techniques have also been explored in an effort to quantify the filtration phenomena, or to visualize flow distributions inside hemodialyzers, these techniques include magnetic resonance imaging (Hardy et al., 2002), computed tomography (Frank et al., 2000, J. C. Kim et al., 2008) and a computerized scanning technique (Ronco et al., 2000, 2002). However, the quantification of internal filtration using these techniques is not available clinically, due to concerns of patient safety and technical requirements.

Summarizing, internal HDF can provide a means of convective treatment by increasing internal filtration rates using specifically designed hemodialyzers, and at the same time spontaneous backfiltration compensates for fluid loss, and hence, this technique is simpler than other modalities. The hemodialyzer design for internal HDF must be optimized based on specified structural factors and on the filtration characteristics of membrane fibers. The literature suggests superior dialysis outcomes for iHDF, but the precise quantification of internal filtration remains to be determined.
