A DIALYSIS APPARATUS WHICH ULTRA-FILTERS BLOOD AND A RELATED METHOD

Abstract

This invention relates to a dialysis apparatus which ultra-filters blood and to a related method of ultra-filtering blood ex-vivo.

More preferably the invention relates to haemodiafiltration using protein-losing membranes and a secondary ultrafiltration and partial re-infusion of haemodiafiltrate for increasing extraction of middle molecules and protein bound uraemic toxins and reducing albumin loss.

Claims

1. A haemodialysis apparatus for ultrafiltering blood from a patient includes: at least a first, at least a second and at least a third pump and, a first dialyser circuit which includes a first haemodialyser with a first haemodialyser membrane which filters blood that is forced by the at least first pump and the at least second pump; and at least third pump and at least a fourth pump which are operative to separate plasma protein bound and unbound uraemic toxins from the blood into a dialysate to form a haemodiafiltrate, characterised in that a second dialyser circuit facilitates dissociation of plasma protein bound uraemic toxins and includes at least a fifth pump and a second haemodialyser with a second haemodialyser membrane which separates uraemic toxins from plasma protein by ultrafiltration to a residue of haemodiafiltrate containing concentrated plasma protein for recirculation to the patient.

2. A dialysis apparatus according to any preceding claim wherein at least a sixth pump is connected to an output of the second haemodialyser and is operative to control the volume flow rate of net fluid removal from the extracorporeal blood circuit (y).

3. A dialysis apparatus according to claim 1 or 2 wherein the second haemodialyser ultra filters at least 90% of the volume of the filtered haemodiafiltrate passing therethrough.

4. A dialysis apparatus according to any preceding claim wherein the second haemodialyser is operative to return at least 3% of protein concentrated haemodiafiltrate to a venous side of the extracorporeal blood circuit.

5. A dialysis apparatus according to any preceding claim wherein the at least third pump is operative to vary the volume flow rate of haemodiafiltrate that is passed to the second haemodialyser.

6. A dialysis apparatus according to any preceding claim wherein the at least a fourth pump is operative to vary a proportion of haemodiafiltrate that is infused to a patient via a venous (return) side of the extracorporeal circuit.

7. A dialysis apparatus according to any preceding claim wherein a pressure source ensures that hydraulic pressure in a section of fluid pathway between the at least third pump and the at least fourth pump is operated within operational limits required for accurate volumetric performance of the at least third pump and the at least fourth pump.

8. A dialysis apparatus according to claim 7 wherein the pressure source is a pump.

9. A dialysis apparatus according to any preceding claim wherein at least a first balance chamber and the at least a second balance chamber and the at least second pump and at least a seventh pump operate to ensure a volume flow rate of HDF replacement fluid, enters the first dialyser circuit for infusion into a venous side of an extracorporeal blood circuit.

10. A dialysis apparatus according to any of claims 1 to 8 wherein the at least first balance chamber is connected to the at least second pump and to the at least fifth pump, whereby the at least first balance chamber ensures correct volume flow rates for the haemodiafiltrate to be infused into the patient in the event of a failure of any pump.

11. A dialysis apparatus according to any preceding claim wherein the at least second balance chamber is connected to the at least first balance chamber and to the at least third pump; and the second balance chamber is operative to ensure that the volume flow rate of haemodiafiltrate is at least equal to the volume flow rate of incoming fresh dialysate.

12. A dialysis apparatus according to any preceding claim wherein the at least fourth pump is operative, in dependence upon operator specified criteria, to vary the ultrafiltration volume flow rate in the first haemodialyser.

13. A dialysis apparatus according to any preceding claim wherein sensors transmit: a signal indicative of hydraulic pressure in the first and the second dialyser circuits; a signal indicative of a pumping pressure; a signal indicative of a pumping speed; a signal indicative of a saline concentration; a signal indicative of an electrical conductance of a solute in haemodiafiltrate; and a signal indicative of concentration of a solute in haemodiafiltrate.

14. A dialysis apparatus according to claim 13 includes a control means to vary a volume of ultrafiltered haemodiafiltrate, passing from the second haemodialyser and to vary a proportion of the protein concentrated haemodiafiltrate for infusion into the venous side of the extracorporeal blood circuit.

15. A dialysis apparatus according to any preceding claim includes a means to vary a volume flow rate of haemodiafiltrate to a venous side of an extracorporeal blood circuit.

16. A dialysis apparatus according to any preceding claim wherein a sensor senses a state of a membrane performance with treatment time and a processor is operative to determine a risk of membrane rupture and is operative to trigger an alarm.

17. A dialysis apparatus according to any preceding claim wherein a sensor is arranged to sense back filtration from the second haemodialyser and is operative to trigger an alarm.

18. A dialysis apparatus according to any preceding claim where the processor is operative to initiate pulses of raised pressure in a dialysate compartment of the first and/or second haemodialysers.

19. A dialysis apparatus according to any preceding claim wherein a filtration membrane is housed within a removable cartridge.

20. A dialysis apparatus according to any preceding claim wherein a monitor is operative to monitor preset patient criteria and to trigger an alarm when a criterion is not satisfied.

21. A dialysis apparatus according to any preceding claim wherein a memory records a patient's details, patient related data, a dialysis start time and a dialysis end time and operational data relating to pumping data.

22. A dialysis apparatus according to claim 21, when dependent on any of claims 13 to 20, wherein a memory records data from the sensors.

23. A method of ultra-filtering human or animal blood using the apparatus according to any preceding claim.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0040] FIG. 1 (PRIOR ART) shows graphs of sieving coefficients against molecular weight of different molecules;

[0041] FIG. 2 is a graph of probability density against membrane pore size and the differential of normalized total membrane ultrafiltration rate against membrane pore size;

[0042] FIG. 3 is a graph of sieving coefficient against molecular size;

[0043] FIG. 4 is a functional block diagram of one embodiment of a dialysis apparatus which ultra-filters blood and to a related method of ultra-filtering blood ex-vivo; and

[0044] FIG. 5 is functional block diagram of a second one embodiment of a dialysis apparatus which ultra-filters blood and to a related method of ultra-filtering blood ex-vivo.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS OF THE INVENTION

[0045] Referring to FIGS. 4 and 5, which each show alternative embodiments of the dialysis apparatus, comprises two dialyser circuits, the first departs from prior art HDF only in that the first haemodialyser may consist of an MCO, HCO or other protein-losing membrane.

[0046] In the second dialyser circuit the effluent t haemodiafiltrate from the first haemodialyser/dialysis circuit passes through the blood compartment of a second MCO or HCO haemodialyser. Ultrafiltration of between 90% and 97% of the volume of haemodiafiltrate passing through the haemodialyser leaves 3% to 10% of protein concentrated haemodiafiltrate for reinfusion into the venous/return side of the extracorporeal blood circuit.

[0047] The proportion of the volume of haemodiafiltrate ultrafiltered in the second haemodialyser (1f) and therefore the proportion infused into the venous side of the extracorporeal blood circuit (f) is a modifiable parameter. The mechanism by this parameter is set differs between the two preferred embodiments of the dialysate/haemodiafiltrate fluid pathway shown in FIGS. 4 and 5 whose control algorithms are as follows:

FIG. 4, First Embodiment Pathway (1)

[0048] 1. Operator set blood pump (P.sub.1) volume flow rate [0049] 2. Operator set first haemodialyser ultrafiltration rate (x) according to blood flow within safety constraints programmed into the processor [0050] 3. Operator set dialysate inflow pump (P.sub.2) volume flow rate (d) [0051] 4. Processor set first haemodialyser HDF pump (P.sub.3)=(d+x) [0052] 5. Operator set f (0.05 in the example shown in FIG. 4) [0053] 6. Processor set second haemodialyser outflow pump (P.sub.4)=0.05(d+x) [0054] 7. Operator set patient net ultrafiltration pump (P.sub.6) volume flow rate (y) [0055] 8. Processor set HDF replacement fluid pump (P.sub.7)=xy0.05(d+x) [0056] 9. Allow pressure sensor driven processor adjustments to P.sub.4, within a certain tolerance, so that flow in the system remains continuous

FIG. 5, Second Embodiment Pathway (2)

[0057] 1. Operator set blood pump (P.sub.1) volume flow rate [0058] 2. Operator set first haemodialyser ultrafiltration rate (x) according to blood flow within safety constraints programmed into the processor [0059] 3. Operator set dialysate inflow pump (P.sub.2) volume flow rate (d) [0060] 4. Processor set first haemodialyser dialysate/ultrafiltration pumps (P.sub.3/P.sub.4) to d and x respectively [0061] 5. Operator set patient net ultrafiltration pump (P.sub.6) volume flow rate (y)

[0062] In the First Embodiment f is preset by the machine operator (0.05 in the example shown in FIG. 4). In the second pathway f is not preset but follows automatically from the operator of the dialysis apparatus choice of x, y and d.

[0063] Both embodiments incorporate at least two balance chambers. These (or an equivalent technical device) are essential for ensuring correct volume flow rates for the dialysate/haemodiafiltrate and to guarantee patient safety in the event of major loss of calibration of any of the pumps. A balance chamber effectively comprises a volume separated into two portions by a membrane. It is appreciated that two pairs of balance chambers operate in tandem.

Theoretical Explanation

[0064] Despite the technical differences between these two Embodiments the same mathematical model can be applied to the analysis of solute extraction. A unidimensional model of the HDF process in the first haemodialyser and the subsequent ultrafiltration of the haemodiafiltrate in the second haemodialyser is presented in the following paragraphs.

[0065] The membrane transfer of solutes, such as albumin, with very low mass transfer coefficients, even across the more permeable types of membrane, may be analysed as being solely due to convection. [0066] If: .sub.x=the volume flow per unit time of the remaining haemodiafiltrate (water volume flow rate for dilute solutions) at a point (x) located along the blood compartment of the second haemodialyser m.sub.x=the mass of solute passing x per unit time [0067] then

[00001]$\frac{m_x}{v_x}=C_x$ =the solute concentration at x

[0068] For present purposes, an expression linking solute extraction to the cumulative level of ultrafiltration, without reference to x, will suffice. The expression can be derived by analysing the solute mass flow rate at positions along the blood compartment of the secondary hemodialysis in terms of m.sub., where represents the remaining haemodiafiltrate volume flow rate.

[0069] As .fwdarw.0 ( being positive in the direction opposite to the flow of haemodiafiltrate)

[00002]$m_{v-v}.fwdarw.m_v-S\left(\frac{m_v}{v}\right)v$$C=C_v-C_{v-v}=\frac{m_v}{v}\left[\frac{\left(S-1\right)v}{v-v}\right]C_v\left[\frac{\left(S-1\right)v}{v}\right]$

where S=the convection coefficient of the solute and:

[00003]$\stackrel{C_{v_i}}{\underset{C_v}{}}\frac{C_v}{C_v}=\left(S-1\right)\stackrel{v_i}{\underset{v}{}}\frac{v}{v}$

where .sub.i is the initial volume flow rate of the haemodiafiltrate in the second haemodialyser

[00004]$\ln C_{v_i}-\ln C_v=\left(S-1\right)\left(\ln v_i-\ln v\right)$$\ln C_v-\ln C_{v_i}=\left(S-1\right)\left(\ln v-\ln v_i\right)$$\frac{C_v}{C_{v_i}}={\left(\frac{v}{v_i}\right)}^{S-1}$

[0070] Letting

[00005]$f=\frac{v}{v_i}$

where f is the proportion or naemodiafiltrate returned to the patient.

[00006]$\frac{C_v}{C_{v_i}}=f^{S-1}\mathrm{Extraction}\left(E\right)=1-f\left(\frac{C_v}{C_{v_i}}\right)=1-f^S$

[0071] As an example, if 50 L of plasma with an albumin concentration of 40 g/l passes through the first haemodialyser with an albumin extraction of 0.01, 20 g of albumin will pass into the hemodiafiltration. If this haemodiafiltrate is then passed through a second haemodialyser in which S for albumin=0.02 and f=0.05, albumin extraction is 0.058 and only 1.16 g of the original 20 g of albumin in the haemodiafiltrate is ultimately lost.

[0072] Analysis of ultrafiltration of hemodiafiltration with regard to the extraction of smaller, more diffusible, LMWP for whom S is less than unity requires a model that takes account of both convection and diffusion.

[0073] The ultrafiltration process in the second haemodialyser may be configured such that the flow of ultrafiltrate runs in a concurrent or countercurrent direction with respect to the flow of the haemodiafiltrate. In either case, given dynamic equilibrium, the sum of the solute fluxes entering and leaving any small segment of the blood compartment between two points (x and x+x) along the active portion of the second haemodialyser must equal zero, where x.sub.0 is the beginning of the active portion and x is positive in the direction of flow of the haemodiafiltrate. Q.sub.x, the haemodiafiltrate volume flow rate, will change with x because of ultrafiltration, as will C.sub.x the concentration of the solute. Qe.sub.x and Ce.sub.x correspond to the volume flow rate and solute concentration respectively in the ultrafiltrate compartment of the second haemodialyser.

[0074] The solute mass conservation equation within the blood compartment of the second haemodialyser at dynamic equilibrium accords with the equation previously derived for prior art HDF: [0075] Letting q.sub.UF(x)=the ultrafiltration volume flow rate per unit length of haemodialyser at x [0076] K=the mass transfer (diffusion) coefficient per unit length of haemodialyser at x [0077] Asx.fwdarw.0:

[00007]$\begin{array}{cc}{Q}_x{C}_x-\left(Q_x-q_{\mathrm{UF}\left(x\right)}x\right)C_{x+x}-q_{\mathrm{UF}\left(x\right)}S\left(\frac{C_x{e}^{\frac{q_{\mathrm{UF}\left(x\right)S}}{K}}-{\mathrm{Ce}}_x}{e^{\frac{q_{\mathrm{UF}\left(x\right)}S}{K}}-1}\right)x0Q_x{C}_x\left(Q_x-q_{\mathrm{UF}\left(x\right)}x\right)\left(C_x+C_x\right)+q_{\mathrm{UF}\left(x\right)}S\left(\frac{C_x{e}^{\frac{q_{\mathrm{UF}\left(x\right)S}}{K}}-{\mathrm{Ce}}_x}{e^{\frac{q_{\mathrm{UF}\left(x\right)}S}{K}}-1}\right)x{C}_x\frac{q_{\mathrm{UF}\left(x\right)}x}{Q_x-q_{\mathrm{UF}\left(x\right)}x}\left[C_x-S\left(\frac{C_x{e}^{\frac{q_{\mathrm{UF}\left(x\right)S}}{K}}-{\mathrm{Ce}}_x}{e^{\frac{q_{\mathrm{UF}\left(x\right)}S}{K}}-1}\right)\right]& 1.\end{array}$

C.sub.x can therefore be derived by numerical integration starting with C.sub.x.sub.o as long as Q.sub.x, q.sub.UF(x) and Ce.sub.x can be defined. When the value of S is less than unity, using C.sub.x to calculate C.sub.x between x and x+x tends to underestimate solute extraction. The degree of error is reduced by splitting the active portion of the haemodialyser into a large number of segments (two hundred in the case of the analysis underlying the results presented in Table 1).

[0078] In the case of concurrent flow, Ce.sub.x can be derived from the boundary conditions:

[00008]$Q_x+{\mathrm{Qe}}_x=Q_{x_o}{Q}_x{C}_x+{\mathrm{Qe}}_x{\mathrm{Ce}}_x=Q_{x_o}{C}_{x_o}$

so that:

[00009]${\mathrm{Ce}}_x=\frac{Q_{x_o}{C}_{x_o}-Q_x{C}_x}{Q_{x_o}-Q_x}$

Q.sub.x and q.sub.UF(x) are related by the fact that

[00010]$q_{\mathrm{UF}\left(x\right)}=-\frac{{\mathrm{dQ}}_x}{\mathrm{dx}}.$

The overall ultrafiltration rate is set by the product of (1f) and the volume flow rate of the haemodiafiltrate. The distribution of ultrafiltration along the length of the second haemodialyser is set by the ratio between two hydraulic coefficients, one governing the volume flow rate of the haemodiafiltrate, the other the ultrafiltration rate across the membrane.

[0079] As the haemodiafiltrate passes through the blood compartment of the second haemodialyser its protein solute content becomes progressively more concentrated. It is assumed ere that this increase in protein concentration has a negligible effect on viscosity and that:

[00011]$\begin{array}{cc}{Q}_x=-k\frac{{\mathrm{dP}}_x}{\mathrm{dx}}& 2.\end{array}$

where P.sub.x=the hydraulic pressure within the blood compartment. (At x.sub.0 a typical albumin concentration within haemodiafiltrate would be of the order of one hundredth of that in the plasma, rising to a little less than a fifth at the end of the active second haemodialyser if f=0.05.)

[0080] The loss of haemodiafiltrate from the blood compartment of the second haemodialyser due to ultrafiltration is given by:

[00012]$\begin{array}{cc}\frac{{\mathrm{dQ}}_x}{\mathrm{dx}}=-h\left(P_x-P_e\right)& 3.\end{array}$

where h is the hydraulic conductivity per unit length of haemodialyser, whose actual numerical value, as with that of k, can come from experimental observation. It is assumed that the ultrafiltrate compartment of the second haemodialyser provides a considerably lower resistance pathway for longitudinal fluid flow than the fibre lumina and therefore that the negative hydraulic pressure (P) required for ultrafiltration remains relatively constant throughout that compartment.

[0081] (In support of this assumption, the technical datasheet accompanying the Baxter Corporation Theranova MCO haemodialyser indicates a pressure drop in the blood compartment with bovine blood and Q.sub.b=300 ml/min of 130 mmHg. The pressure drop in the dialysate compartment with Q.sub.D=500 ml/min is quoted as 30 mmHg.)

[0082] From equation 3:

[00013]$\frac{d^2{Q}_x}{{\mathrm{dx}}^2}=-h\frac{{\mathrm{dP}}_x}{\mathrm{dx}}$

[0083] Combining with equation 2:

[00014]$\frac{d^2{Q}_x}{{\mathrm{dx}}^2}=\frac{h}{k}{Q}_x$

[0084] This has a general solution:

[00015]$Q_x={\mathrm{Ae}}^{-{\left(\frac{h}{k}\right)}^{\frac{1}{2}}x}+{\mathrm{Be}}^{{\left(\frac{h}{k}\right)}^{\frac{1}{2}}x}$

[0085] A and B can be determined from the following boundary conditions: [0086] Q.sub.x=Q.sub.x.sub.0 when x=x.sub.0 [0087] Q.sub.x.sub.L=fQ.sub.x.sub.0 when x=L, where L=the length of the active portion of the haemodialyser

A+B=Q.sub.x.sub.0

therefore:

[00016]$A=Q_{x_o}\left[\frac{e^{{\left(\frac{h}{k}\right)}^{\frac{1}{2}}L}-f}{e^{{\left(\frac{h}{k}\right)}^{\frac{1}{2}}L}-e^{-{\left(\frac{h}{k}\right)}^{\frac{1}{2}}L}}\right]\mathrm{and}B=Q_{x_o}\left[\frac{f-e^{-{\left(\frac{h}{k}\right)}^{\frac{1}{2}}L}}{e^{{\left(\frac{h}{k}\right)}^{\frac{1}{2}}L}-e^{-{\left(\frac{h}{k}\right)}^{\frac{1}{2}}L}}\right]$

[0088] Numerical integration of equation 1 provides the value for C.sub.x.sub.L and the extraction of solute from haemodiafiltrate=

[00017]$E=1-f\frac{C_{x_L}}{C_{x_o}}$

[0089] For countercurrent flow the boundary conditions for mass conservation within the haemodialyser are:

[00018]$C_x{Q}_x-C_{x_L}{Q}_{x_L}={\mathrm{Ce}}_x{\mathrm{Qe}}_x{\mathrm{Ce}}_{x_o}{\mathrm{Qe}}_{x_o}+C_{x_L}{Q}_{x_L}=C_{x_o}{Q}_{x_o}{C}_{x_o}{Q}_{x_o}-C_x{Q}_x={\mathrm{Ce}}_{x_o}{\mathrm{Qe}}_{x_o}-{\mathrm{Ce}}_x{\mathrm{Qe}}_x$

[0090] Combining the first and second of these boundary conditions and then subtracting the third gives:

[00019]${\mathrm{Ce}}_x=\frac{C_x{Q}_x-C_{x_L}{Q}_{x_L}}{Q_x-Q_{x_L}}$

[0091] This can be used to provide values of Ce.sub.x for numerical integration given a certain value for C.sub.x.sub.L. An iterative process is therefore required which, in achieving internal consistency as to C.sub.x.sub.L, establishes its correct value. For larger LMWP, whose extraction is the main focus of the dialysis apparatus and for whom convection is a more important mechanism, calculations show that whether the haemodiafiltrate and ultrafiltrate are directed to run in concurrent or countercurrent directions makes very little difference to the extraction values produced by the model.

Predictions of in Vivo Solute Extraction

[0092] To illustrate the impact of this technique on solute extraction using MCO and/or HCO haemodialysers, values of K.sub.oA and S, taken to be representative of four molecular size classes, were tested using the mathematical model. The four classes of molecule were: small solutes (for example phosphate), smaller LMWP (e.g. 2 microglobulin), larger LMWP (e.g. interleukin 6) and large plasma proteins (e.g. albumin). The values of K.sub.oA and S assigned to the various haemodialyser/solute molecular class combinations were estimated using the product information sheets and promotional literature pertaining to the Nippro Elisio High Flux haemodialyser, the Theranova MCO haemodialyser (Baxter Healthcare Corporation) and the SepteX HCO haemodialyser (Gambro-Baxter, Hechingen, Germany) and the results of an earlier study.

[0093] For this illustration the direction of flow of the haemodiafiltrate in the second haemodialyser is taken to be concurrent with that of the ultrafiltrate. For all four classes of molecule, it has been assumed that the effective distribution volume, at least for the duration of a single transit of the haemodialyser, is the plasma water.

[0094] Example volume flow rate parameters chosen for the illustration were a plasma water volume flow rate (Q.sub.b.sub.e) of 200 ml/min, dialysate volume flow rates (d) of 600 ml/minor 1200 ml/min and a primary haemodialyser haemofiltration rate (x) of 70 or 80 ml/min. The internal filtration rates (IFR) of the three types of haemodialyser were set at different levels to reflect their differing hydraulic conductivities. The patient net ultrafiltration rate was set to either zero or 10 ml/min as shown.

[0095] Three different combinations of these volume flow rate parameters were studied using the model. The results shown in Table 1 were calculated on the assumption that Pathway 1 was operative whereas in Tables 2 and 3 Pathway 2 was operative.

[0096] In the absence of actual experimental data, as a guide to the relationship between Q.sub.x and distance (x) along the second haemodialyser, estimates of the two hydraulic coefficients k and h (see above) were derived as follows:

[0097] Assuming a distance-averaged volume flow rate of circa 300 cm.sup.3/min in the blood compartment and an overall pressure gradient of 100 mmHg along a haemodialyser of 20 cm active length, equation 2 was used to provide an estimate of k=60 cm.sup.4/mmHgmin.

[0098] Assuming an average ultrafiltration rate of 30 cm.sup.3/min per cm of active haemodialyser with a uniform negative dialysate compartment pressure of 300 mmHg and an average positive pressure within the fibre lumina of +50 mmHg, equation 3 was used to provide an estimate of h=(30/350) cm.sup.2/mmHg min.

TABLE-US-00001 TABLE 1 (Pathway 1) A comparison of the extraction of solutes of differing molecular weight by HDF using High Flux (HF), Medium Cut Off (MCO) and High Cut Off (HCO) haemodialysers (first three columns) and (last three columns) the dialysis apparatus utilises Dialysate Pathway 1 (primary and secondary haemodialysers as shown, 95% ultrafiltration of haemodiafiltrate, reinfusion of the residual 5%). E.sub.1 = solute extraction by the primary haemodialyser, E.sub.2 = solute extraction by ultrafiltration of the haemodiafiltrate in the secondary haemodialyser, E.sub.1, E.sub.2 = overall solute extraction. K.sub.oA = mass transfer area coefficient (ml/min), S = convection coefficient, Q.sub.b.sub.e = plasma volume flow rate, d = volume flow rate of fresh dialysate, x = primary haemodialyser haemofiltration rate, IFR = internal filtration rate. Q.sub.b.sub.e = 200 ml/min d = 500 ml/min HCO primary x = 70 ml/min Single HF Single MCO Single HCO with MCO f = 0.05 IFR = 20 ml/min IFR = 30 ml/min IFR = 40 ml/min Dual MCO Dual HCO secondary (net UF = 0) K.sub.oA S E.sub.1 K.sub.oA S E.sub.1 K.sub.oA S E.sub.1 E.sub.2 E.sub.1, E.sub.2 E.sub.2 E.sub.1, E.sub.2 E.sub.2 E.sub.1, E.sub.2 Small molecule 500 1 0.953 700 1 0.986 1000 1 0.998 0.95 0.937 0.95 0.948 0.95 0.948 (e.g. PO.sub.4) (98% of (99% of (99% of single HF) single HF) single HF) Smaller LMWP 40 0.6 0.351 80 1 0.559 160 1 0.745 0.95 0.531 0.95 0.708 0.95 0.708 (e.g. 2M) (151% of (202% of (202% of single HF) single HF) single HF) Larger LMWP 2 0.2 0.089 6 0.6 0.256 18 0.7 0.342 0.833 0.213 0.876 0.300 0.833 0.285 (e.g. IL6) (239% of (337% of (320% of single HF) single HF) single HF) Large plasma 0 0.002 0.00092 0.2 0.02 0.0102 0.6 0.053 0.0304 0.057 0.00058 0.145 0.0044 0.057 0.00173 proteins e.g. (63% of (478% of (188% of albumin single HF) single HF) single HF) Expected albumin 1.77 grams 19.6 grams 58.4 grams 1.11 grams 8.45 grams 3.32 grams loss per session assuming 48 litres of plasma processed and average [albumin] 40 g/l

TABLE-US-00002 TABLE 2 (Pathway 2) A comparison of the extraction of solutes of differing molecular weight by HDF using high flux (HF), medium cut off (MCO) and high cut off (HCO) haemodialysers (first three columns) and (last three columns) the dialysis apparatus n incorporating Dialysate Pathway2 (primary and secondary haemodialysers as shown, 89.7% ultrafiltration of haemodiafiltrate, reinfusion of the residual 10.3%). E.sub.1 = solute extraction by the primary haemodialyser, E.sub.2 = solute extraction by ultrafiltration of the haemodiafiltrate in the secondary haemodialyser, E.sub.1, E.sub.2 = overall solute extraction. K.sub.oA = mass transfer area coefficient (ml/min), S = convection coefficient, Q.sub.b.sub.e = plasma volume flow rate, d = volume flow rate of fresh dialysate, x = primary haemodialyser haemofiltration rate, IFR = internal filtration rate. Q.sub.b.sub.e = 200 ml/min d = 600 ml/min x = 80 ml/min Single HF Single MCO Single HCO HCO primary with f = 0.103 IFR = 20 ml/min IFR = 30 ml/min IFR = 40 ml/min Dual MCO Dual HCO MCO secondary net UF = 10 ml/min K.sub.oA S E.sub.1 K.sub.oA S E.sub.1 K.sub.oA S E.sub.1 E.sub.2 E.sub.1, E.sub.2 E.sub.2 E.sub.1, E.sub.2 E.sub.2 E.sub.1, E.sub.2 Small molecule 500 1 0.960 700 1 0.989 1000 1 0.999 0.90 0.890 0.90 0.899 0.90 0.899 (e.g. PO.sub.4) (93% of (94% of (94% of single HF) single HF) single HF) Smaller LMWP 40 0.6 0.379 80 1 0.590 160 1 0.769 0.90 0.531 0.90 0.692 0.90 0.692 (e.g. 2M) (140% of (183% of (183% of single HF) single HF) single HF) Larger LMWP 2 0.2 0.104 6 0.6 0.287 18 0.7 0.374 0.743 0.213 0.796 0.298 0.743 0.278 (e.g. IL6) (205% of (287% of (267% of single HF) single HF) single HF) Large plasma 0 0.002 0.0011 0.2 0.02 0.0117 0.6 0.053 0.0342 0.044 0.0005 0.113 0.0039 0.044 0.0015 proteins e.g. (45% of (355% of (136% of albumin single HF) single HF) single HF) Expected albumin 2.11 grams 22.4 grams 65.7 grams 0.96 grams 7.49 grams 2.88 grams loss per session assuming 48 litres of plasma processed and average [albumin] 40 g/l

TABLE-US-00003 TABLE 3 (Pathway 2) A comparison of the extraction of solutes of differing molecular weight by HDF using High Flux (HF), Medium Cut Off (MCO) and High Cut Off (HCO) haemodialysers (first three columns) and (last three columns) the dialysis apparatus incorporates Dialysate Pathway 2 (primary and secondary haemodialysers as shown, 94.5% ultrafiltration of haemodiafiltrate, reinfusion of the residual 5.5%). E.sub.1 = solute extraction by the primary haemodialyser, E.sub.2 = solute extraction by ultrafiltration of the haemodiafiltrate in the secondary haemodialyser, E.sub.1, E.sub.2 = overall solute extraction. K.sub.oA = mass transfer area coefficient (ml/min), S = convection coefficient, Q.sub.b.sub.e = plasma volume flow rate, d = volume flow rate of fresh dialysate, x = primary haemodialyser haemofiltration rate, IFA = internal filtration rate. Q.sub.b.sub.e = 200 ml/min d = 1200 ml/min HCO primary x = 80 ml/min Single HF Single MCO Single HCO with MCO f = 0.055 IFR = 20 ml/min IFR = 30 ml/min IFR = 40 ml/min Dual MCO Dual HCO secondary net UF = 10 ml/min K.sub.oA S E.sub.1 K.sub.oA S E.sub.1 K.sub.oA S E.sub.1 E.sub.2 E.sub.1, E.sub.2 E.sub.2 E.sub.1, E.sub.2 E.sub.2 E.sub.1, E.sub.2 Small molecule 500 1 0.969 700 1 0.993 1000 1 0.999 0.945 0.938 0.945 0.944 0.945 0.944 (e.g. PO.sub.4) (97% of (97% of (97% of single HF) single HF) single HF) Smaller LMWP 40 0.6 0.377 80 1 0.594 160 1 0.777 0.945 0.561 0.945 0.734 0.945 0.734 (e.g. 2M) (149% of (195% of (195% of single HF) single HF) single HF) Larger LMWP 2 0.2 0.101 6 0.6 0.284 18 0.7 0.371 0.823 0.234 0.868 0.322 0.823 0.305 (e.g. IL6) (232% of (319% of (302% of single HF) single HF) single HF) Large plasma 0 0.002 0.00106 0.2 0.02 0.0114 0.6 0.053 0.033 0.056 0.00064 0.141 0.0047 0.056 0.00185 proteins e.g. (60% of (443% of (175% of albumin single HF) single HF) single HF) Expected albumin 2.04 grams 21.9 grams 63.4 grams 1.23 grams 9.02 grams 3.55 grams loss per session assuming 48 litres of plasma processed and average [albumin] 40 g/l

[0099] These results show that it is possible to increase the extraction of smaller and larger LMWP to varying degrees (with increases of between 51% and 237% in the examples given in Tables 1, 2 and 3) using the dialysis apparatus and to achieve those increases without excessive loss of albumin. The Second Embodiment (Pathway 2) requires twice the volume flow rate of fresh dialysate to achieve similar levels of LMWP as the First Embodiment (Pathway 1) making the latter the likely preferred option from an economic and sustainability perspective.

[0100] Although the normal physiological rate of renal albumin loss is only 600 mg per day, daily turnover of albumin is approximately 10 grams. On this basis a thrice weekly albumin loss from therapy of 8.45 grams using two HCO haemodialysers. The dialysis apparatus might be considered reasonable, however the option that appears to provide the best compromise between higher LMWP extraction and minimal albumin loss involves the use of an HCO membrane as the primary haemodialyser and an MCO membrane as the secondary haemodialyser. Under these circumstances, assuming 95% ultrafiltration of haemodiafiltrate with Pathway 1, a minor reduction (1-2%) in the extraction from plasma water of small solutes compared to prior art treatment may be anticipated, however the impact of this reduction on aggregate small solute removal over the entire course of a dialysis treatment session will be marginal, inter compartmental solute kinetics being a far more important limiting factor.

[0101] The invention may also be deployed to remove protein bound uraemic toxins (PBUT), such as indoxyl sulphate (IS) and p-cresylsulphate (pCS) which are small solutes actively secreted from the peritubular capillary blood of healthy kidneys into the lumen of the proximal tubule. Albumin is the predominant binding protein for these toxins leaving a free fraction of circa 0.1 for IS and 0.07 for pCS. Albumin binding significantly reduces the efficiency of IS and pCS extraction by prior art high flux membrane haemodialysis with or without high volume ultrafiltration. The predominant mechanism(s) by which albumin binding reduces PBUT extraction is uncertain but the conclusion of the analysis presented below is that the effect on volume of distribution (V.sub.D) is unlikely to be the sole cause of the modest post treatment reductions in PBUT concentration (20% to 30%) that are observed.

[0102] If it is assumed that during hemodialysis treatment PBUT are effectively restricted to the plasma compartment and that the kinetics of their dissociation from albumin and the kinetics of their mass transfer, including those between the blood and dialysate compartments of the haemodialyser, are not rate limiting then their reduction ratio following a session of haemodialysis treatment can be estimated as follows:

[0103] Assuming a total blood volume of 5 L and a haematocrit of 30% the total plasma volume will be 3.5 L. If the free fraction of the PBUT for example CS is taken to be 0.07 its V.sub.D will be approximately:

[00020]$3.5\left(1+\frac{93}{7}\right)=50L$

[0104] If the blood volume flow rate to the haemodialyser were 300 ml/min i.e. 210 ml/min in terms of plasma flow, the dialysate flow rate were 600 ml/min and PBUT mass transfer kinetics were non rate limiting, the extraction of PBUT by the haemodialyser would be:

[00021]$\frac{600}{600+200\left(1+\frac{93}{7}\right)}=0.167$

[0105] The above calculation also depends upon the assumption that serum albumin, as part of the V.sub.D of the PBUT, constitutes a non-saturable compartment, at least over the range of PBUT concentrations found in practice. In the absence of any definitive data with respect to albumin PBUT binding capacity, and binding affinities, it is acknowledged that this assumption can only be seen as a first approximation. Assuming single compartment kinetics the equation governing the extraction of PBUT from plasma is therefore:

[00022]$\frac{d\left[\mathrm{PBUT}\right]}{\mathrm{dt}}=\frac{-0.167201\left(1+\frac{93}{7}\right)}{50,000}\left[\mathrm{PBUT}\right]=-0.01\left[\mathrm{PBUT}\right]$

[0106] If a dialysis session were of 240 minutes duration it follows that:

[00023]$\frac{\left[\mathrm{PBUT}\right]}{{\left[\mathrm{PBUT}\right]}_0}=0.09$

[0107] That is a 91% reduction ratio (RR).

[0108] As this level of reduction ratio is not observed in practice using prior art haemodialysis and HDF techniques the assumptions that the effective V.sub.D of PBUT is restricted to the plasma compartment and/or that the kinetics of PBUT albumin dissociation and their mass transfer kinetics, including that across the haemodialyser, are not limiting factors, must come into question. Approximately 40% of total extracellular albumin is found in the plasma compartment with the remaining 60% in the interstitial fluid. If this extra V.sub.D is included in the reduction ratio calculation (with an additional 10 L to allow for the expansion of the interstitial fluid compartment in dialysis patients) this would make a total V.sub.D of 135 L.

[0109] Even with this adjustment the predicted single compartment kinetics reduction ratio becomes 59%, still well in excess of what is seen using prior art haemodialysis related techniques. One published mathematical model of PBUT kinetics during haemodialysis includes the intracellular water as part of the V.sub.D for PBUT. This would add approximately another 30 L to the V.sub.D but the V.sub.D so created is very unlikely to function as a uniform compartment. It is also somewhat counterintuitive to assume that a potentially toxic metabolite such as PCS, secreted by the liver and, in renal health, destined for secretion in the proximal tubule, would also be taken up by other cells in the body thereby reducing the efficiency of its elimination.

[0110] Published rate constants for the dissociation of PBUT are not readily available however the binding interaction appears to be relatively weak, reliant upon van der Waals forces, which suggests that the dissociation rate constants are likely to be quite high. The focus is therefore on the kinetics of PBUT diffusive mass transfer including that from the blood to the dialysate compartments of the haemodialyser. Their albumin binding suggests that the diffusive transfer of PBUT may also be impaired by interactions with other large proteins.

[0111] For example, if the first haemodialyser has an albumin sieving coefficient of 0.1 and the haemofiltration rate were 70 ml/min with a plasma flow rate of 210 ml/min then, considering only PBUT dissociation from albumin in the haemodiafiltrate and ignoring diffusive flux of unbound PBUT across the haemodialyser membrane, PBUT clearance into the primary haemodiafiltrate would be:

[00024]$70+\left(700.1\frac{0.93}{0.07}\right)=163\mathrm{ml}/\min$

[0112] The volume into which these PBUT are then distributed is expanded by a dialysate flow of 600 ml/min and therefore, assuming non saturable PBUT albumin binding, the proportion found free in the haemodiafiltrate will be:

[00025]$\frac{670}{670+93}=0.88$

[0113] If 95% of the volume of haemodiafiltrate is removed by the secondary ultrafiltration process and all of the albumin is returned, the proportion of unbound PBUT in haemodiafiltrate extracted by the secondary ultrafiltration process is:

[00026]$0.880.95=0.836$

the resulting clearance of PBUT is:

[00027]$0.836163\mathrm{ml}/\min =136\mathrm{ml}/\min$

as a proportion of total V.sub.D (including interstitial albumin) the removal rate is:

[00028]$\frac{136}{135,000}=1{10}^{-3}\mathrm{per}\mathrm{minute}$

[0114] Over the 240 minutes of a treatment session the resulting contribution to the PBUT reduction ratio would be approximately 21%, exclusive of any components due to diffusive flux and/or albumin losses across the second haemodialyser.

[0115] Similar calculations based on the flow parameters shown in Tables 2 and 3, indicate that increasing the flow of fresh dialysate (d) from 600 to 1200 ml/min only increases the reduction in post sessional PBUT concentration from 23% to 25% (again ignoring diffusive flux and losses associated with albumin).

[0116] Greater increases in post sessional PBUT concentration reduction can be achieved by employing first haemodialysers with higher albumin sieving coefficients. Based on the volume flow rate parameters shown in Table 1 but using a first haemodialyser with an albumin sieving coefficient of unity the expected reduction would be 51% increasing to 65% if the dialysate volume flow rate is increased to 1200 ml/min. The use of very porous membranes in the first haemodialyser will tend to increase albumin losses. However, if the potential for increasing PBUT extraction is seen as the major therapeutic advantage of the invention then the use of less porous membranes, for example standard high flux haemodialysers, as the second hemodialysis in combination with a highly porous first haemodialyser would allow high levels of PBUT extraction to be achieved whilst keeping albumin losses to a minimum albeit sacrificing the increase in LMWP extraction that could otherwise be achieved

[0117] The invention has been described by way of examples only and it will be appreciated that variation may be made to the aforementioned embodiments, without departing from the scope of protection, as defined by the claims.