Nephrol Dial Transplant (2003) 18 [Suppl 7]: vii10–vii20
DOI: 10.1093/ndt/gfg1073
Evolution of synthetic membranes for blood purification:
the case of the Polyflux family
Claudio Ronco, Carlo Crepaldi, Alessandra Brendolan, Luisa Bragantini, Vincent d’Intini,
Paola Inguaggiato, Monica Bonello, Bernd Krause1, Reinhold Deppisch1, Herman Goehl1 and
Andrea Scabardi
Department of Nephrology, St Bortolo Hospital, Viale Rodolfi, 36100 Vicenza, Italy and 1Gambro Dialysatoren GmbH
and Co. KG, Corporate Research, Holger-Crafoord-Str. 26, D-72379 Hechingen, Germany
Introduction
Transport in biological systems is controlled by
membranes functioning as barriers between compart-
ments with various degrees of specificity. In 1972,
Singer and Nicolson [1] introduced a breakthrough view
on structures of cell membranes with their fluid mosaic
model. Their findings emerged along colloid chemistry-
based organization of hydrophilic–hydrophobic poly-
mer molecules and structures to form the cell
membrane. Obviously, these principles also influenced
the design of artificial membranes for therapeutic
purposes. The general functional features involved in
membrane separation processes, whether a natural cell
membrane or an artificial membrane, concern either the
process or the structure of the barrier, i.e. diffusion and
convection of solutes along a pressure or concentration
gradient, adsorption to generate a barrier for certain
molecules, patch-like or mosaic-like structures to
prevent or limit unspecific interaction, short transport
channels for fast transport, nano- and micro-dimen-
sions to suppress unspecific interaction and enable
selectivity, narrow pore distribution to sharpen cut-off
profiles, and optimized cross-filtration flux. In bio-
logical organisms, a number of additional functional
elements of membranes are applied: active epithelial
layers supported by basement membrane structures
such as glomerular membranes, confluent tightly
connected layers of endothelial cells, and strictly
ordered hydrophilic–hydrophobic molecular arrange-
ments in cellular bi-layer membranes. The function of
the artificial kidney has relied from the beginning on
principles laid down in renal physiology or, later, in
biophysical principles in synthetic membranes whose
functions were approaching those of the human
Correspondence and offprint requests to: Claudio Ronco,
Department of Nephrology, St Bortolo Hospital, Viale Rodolfi,
36100 Vicenza, Italy. E-mail: cronco@goldnet.it
ß 2003 European Renal Association–European Dialysis and Transplant Association
Downloaded from https://academic.oup.com/ndt/article/18/suppl_7/vii10/1818267 by guest on 16 April 2024
glomerular membrane. Diffusion and convection
predominantly have been combined to achieve the
desired blood purification.
Today, haemodialysis is basically divided into two
categories: blood purification by low-flux and high-flux
dialysis membranes [2–13]. Low-flux dialysis includes
the standard haemodialysis technique in which dialy-
sers with low hydraulic permeability are utilized. Blood
and dialysate flows are at 250 and 500 ml/min,
respectively, and the average duration is 4 h. When
dialyser surface area is increased, blood and dialysate
flows are increased up to 400 and 800 ml/min, respec-
tively, and the treatment is defined in these circum-
stances as high efficiency haemodialysis. In this
treatment, the clearance for small molecules is remark-
ably increased and the treatment time can generally be
shortened, but the clearance for medium to large
molecules is still very low due to the sieving charac-
teristics of the membrane [14–30].
Classic cellulosic membranes such as Cuprophan
were typically hydrogel-type low-flux membranes (i.e.
permeability to water in the range of 5–6 ml/h

mmHg
m2), strongly hydrophilic and remarkably
thin (6–12 mm) to permit an optimal utilization in
diffusive transport of water-soluble solutes for treat-
ments such as haemodialysis.
On the other hand, classic high-flux synthetic
membranes were originally fully hydrophobic, strongly
asymmetric and with a wall thickness of 40–60 mm.
They were originally employed in convective therapies
such as haemofiltration [31] due to their high hydraulic
permeability (i.e. 30–40 ml/h
mmHg
m2) and their
elevated sieving coefficients.
In a subsequent step, membrane structures have been
adapted to fulfil the demand for increased diffusive
permeability matched with tailored convective proper-
ties. This was achieved by (i) using polymer blends of
hydrophilic and hydrophobic polymers; (ii) reduction
of the wall thickness; and (iii) structural modifications
The Polyflux family of synthetic membranes
of the membrane morphology and surface. These
modifications have permitted their application in new
techniques such as high-flux haemodialysis and haemo-
diafiltration [32–35] in which diffusion and convection
are conveniently and efficiently combined.
When high flux membranes are utilized, three
different approaches can be chosen.
(i) Haemofiltration is a fully convective treatment in
which large amounts of ultrafiltrate are produced
and replaced by an ultrapure substitution fluid
[36], and no dialysis fluid is present. This technique
offers the advantage of excellent clearances for
large molecules. The treatment, however, has only
been clinically applied in Europe because of the
need for large quantities of substitution fluid in
commercially prepared bags or in continuous
therapies for acute renal failure (e.g. CVVH).
(ii) Haemodiafiltration is a form of therapy in which
diffusive and convective transport occurs simulta-
neously. Dialysate is circulated countercurrent to
blood, but the amount of ultrafiltration exceeds
the programmed patient weight loss, and therefore
the final fluid balance is achieved by the infusion
of appropriate amounts of substitution fluid in
the arterial or venous line [37–39]. This allows
different modes, i.e. pre- or post-dilution. Due to
high costs related to replacement solutions, new
approaches have been proposed in Europe for
performing haemofiltration and haemodiafiltra-
tion [40–42]. Part of the fresh dialysate is diverted
from its line to the dialyser, is filtered and finally
used as an ultrapure replacement solution. These
on-line treatments have the advantage of reduced
costs, unlimited production of substitution fluid
and no need for commercially prepared fluids in
bags. On the other hand, the microbiological and
chemical purity of the on-line produced fluid may
represent a concern since no on-line controls can
be made; however, advanced bioanalytical assess-
ment and increasing clinical experience indicates
high fluid quality of on-line processed substitu-
tion/infusion fluids.
(iii) High-flux dialysis is a technique in which highly
permeable dialysers are utilized in conjunction
with an accurate ultrafiltration control by the
dialysis machine. Because of the nature of the
hollow fibre dialyser, the correct fluid balance is
achieved thanks to an internal filtration in the
‘proximal’ part and a backfiltration in the ‘distal’
part of the filter [43]. As a consequence, some of
the advantages of removing larger molecules are
maintained because of the internal convection,
while the need for replacement solution is avoided
by a significant amount of backfiltration of fresh
dialysate in the ‘distal’ part. In this way, some of
the advantages of haemodiafiltration are main-
tained with a simpler and easier layout of the
technique [43]. Again, the dialysate purity is a key
issue in this treatment where bacterial products
could be backtransported into the blood together
vii11
with the reverse flux of water. This may affect the
entire concept of biocompatibility of the system.
However, synthetic membranes prepared from
hydrophilic–hydrophobic blends have been shown
to act in contrast to cellulosic materials as an
adsorptive barrier against bacterially derived
products from the dialysate. Concerning this
aspect, great efforts have been made in improving
the quality of the polymeric materials to better
serve biological requirements [44,45] needed for
dialysis membranes and at the same time of the
water treatment systems [46,47].
Downloaded from https://academic.oup.com/ndt/article/18/suppl_7/vii10/1818267 by guest on 16 April 2024
Structure and composition
From basic mass transport equations [48], key
structural membrane features affecting convective and
diffusive transport of solutes from blood into the
dialysate compartment can be identified. The most
relevant membrane parameters are pore radius, surface
porosity and membrane thickness. Besides these, a
number of other parameters control mass transport
through the membrane. The sieving coefficients dur-
ing convective therapies of a membrane are governed
substantially by the number of pores, their size and
distribution in the innermost selective membrane layer.
Membranes can be divided into different categories,
i.e. according to their chemical composition, structural
design and pore conformation. The final membrane
morphology, their hydraulic and diffusive permeabil-
ity, sieving characteristics and fluid interaction (at the
blood–membrane and dialysate–membrane interface)
depend on the manufacturing techniques applied as
well as the polymer composition chosen. Nearly all
dialysis membranes are prepared by diffusion-induced
phase separation. Here the solute sieving coefficients
are tailored by the speed/kinetics of the phase separa-
tion and the different process parameters, e.g. tempera-
ture, polymer composition and precipitation fluid
composition. Under different conditions, the geo-
metrical structure of the membrane can become finely
sponge-like or finger-type, and the porous structure
can become nano- or micro-porous. The membrane
formation process also allows creation of different
layers in the membrane cross-section with specific
functional, and therefore advantageous, clinical prop-
erties: (i) the internal skin layer providing a selective
sieving barrier for solutes; (ii) the inner structure offer-
ing mechanical strength and resistance/stabilization of
the pore conformation during sterilization procedures;
and (iii) the external layer with barrier characteristics
for endotoxin retention/adsorption (Figure 1).
The case of the Polyflux membrane
The Polyflux membrane family represents a series of
membranes originally derived from the synthetic
polymer polyamide. The ‘parent’ membrane was
developed in the late 1970s/early 1980s to serve best
vii12 C. Ronco et al.

Downloaded from https://academic.oup.com/ndt/article/18/suppl_7/vii10/1818267 by guest on 16 April 2024

Fig. 1. Scanning electron micrograph of the cross-section of a Polyflux S membrane and further magnifications of the inner selective layer
showing the actual granular separation structure. The cross-section of a Polyflux (high-flux) membrane clearly allows identification of its novel
three-layer structure. The three layers are defined as follows. (i) Inner selective layer: shows a narrow pore size distribution and is responsible
for the separation of different molecules based on size exclusion. The narrow pore size distribution is responsible for the selectivity of the
separation. This layer has a thickness of 0.1–0.5 mm. (ii) Support layer: the inner selective layer is supported by a more open membrane
structure, which has the function of stabilizing the membrane. This layer has no sieving function and a thickness of 2–5 mm. (iii) Finger-type
structure: layers (i) and (ii) are surrounded by a very open ‘finger-type structure’, which gives additional mechanical stability. Due to the high
void fraction of this layer, the diffusive resistance in this part is identical to that in water. This layer has a thickness of 40–45 mm and presents
an external skin with hydrophobic domains.

in the requirements for haemofiltration therapy. A
strongly asymmetric membrane structure was chosen.
In the next step of evolution of the Polyflux mem-
brane family, diffusive transport characteristics were
enhanced to enable haemodiafiltration, by blending the
hydrophobic polymer with hydrophilic polymer sys-
tems [34,49,50]. The challenging problem in manufac-
turing such membranes was to make reasonable
predictions early in development for the selection of
efficient technologies to allow high precision and large-
scale technology, including full device sterilization after
packaging and suitable device designs.
The strategies involved in/applied to the membrane
optimization include (i) the selection of polymer
materials compatible with being blended, and resistant
to mechanical stress and sterilization procedures;
(ii) definition of surface characteristics providing
hydrophilic–hydrophobic domains in the range below
50 nm [51,52] to balance minimal activation of blood
components; (iii) a membrane formation process
creating an asymmetric three-layer structure enabling
a selective functional barrier and maximal permeabil-
ity; and (iv) the application of specific process
technology for precise production and accuracy of the
micro- as well as the nano-structure design, i.e. pore
size and conformation.
The selection of suitable polymer systems must cope
with the increasing clinical requirements, including
sterilization methods as alternatives to ETO, i.e. steam
and gamma; permeability for middle molecules, but
preventing albumin loss for optimized selectivity in
high volume exchange therapy modes; limiting comple-
ment and subsequent white blood cell activation; and
minimal induction of the coagulation cascade.
The Polyflux membrane [34,49,50] is a blend of
polyamide [53], which provides endotoxin retention
[54,55] due to the hydrophobic sites and improved
biocompatibility due to minimal interaction with blood
components; polyarylethersulfone, which provides
mechanical strength and resistance to heat steriliza-
tion; and finally PVP, i.e. polyvinylpyrrolidone fre-
quently applied in pharmaceutical formulations, which
The Polyflux family of synthetic membranes
Fig. 2. The Polyflux membrane results from a blend of chemical components prevalently present in the different layers of the membrane wall.
contributes to the hydrophilic domains in the surface
and the enhanced diffusive permeability (Figure 2).
Separation performance of a membrane strongly
depends on (i) the structural characteristics of the active
separation layer; (ii) the overall morphology of the
membrane wall; and (iii) the chemical composition of
the membrane, i.e. polymeric materials (blend) used to
manufacture the membrane and their distribution
across the hollow fibre geometry. To allow optimized
mass transport and a high selectivity, a unique structure
is required. In this context, the basic design reasoning
was to opt for a strongly asymmetric three-layer
structure providing the smallest pore size directly at the
interface with the blood compartment (Figure 1). For
membranes used in extracorporeal devices for blood
purification, an additional set of performance char-
acteristics is required, i.e. low protein adsorption as an
intrinsic condition to reduce unspecific and unwanted
blood–membrane interaction, and enabling high
biocompatibility from a general perspective. The
Polyflux membrane family, irrespective of whether
tailored to high- or low-flux properties, has a typical
three-layer structure, as depicted in the scanning
electron micrograph in Figure 2, which is designed to
allow optimized convective and diffusive mass trans-
port in combination with an excellent biocompatibility.
vii13
Downloaded from https://academic.oup.com/ndt/article/18/suppl_7/vii10/1818267 by guest on 16 April 2024
The membrane formation process allows tailoring of
the morphology of the membrane wall, which has an
effect on the convective performances of the mem-
brane. In particular, it is important to analyse the
specific composition of each layer in relation to the
polymeric composition of the blend. If a greater
fraction of PVP is present, the hydrophilic nature of the
membrane might be enhanced, and so is diffusion. This
is indeed the case for the inner skin layer of the Polyflux
membrane where a tailored composition contributes to
an increased permeability to middle molecular weight
solutes and a remarkable selectivity with restriction of
the passage of albumin. If polyamide prevails in the
studied layer, hydrophobic domains will contribute
to the adsorption characteristics and the creation of
a functional barrier to endotoxin from the potential
bacterial dialysate contamination (Figure 3).
In all these considerations, in fact, one important
feature of the membrane is its interaction with water.
Original high-flux membranes were almost completely
hydrophobic and this caused some unwanted effects in
terms of protein interaction and low diffusivity.
Modern synthetic membranes have been modified
so as to achieve a higher degree of interaction with
water and more hydrophilic characteristics. This was
achieved by addition of PVP as a strongly hydrophilic
vii14
Fig. 3. Composition analysis of the Polyflux fibre with ESCA (electron spectroscopy for chemical analysis) and NMR (nuclear magnetic
resonance).
Fig. 4. Atomic force micrographs (tapping mode) of the inner surface of a Polyflux S membrane. Scan size: 5  5 mm.
component to the polymeric blend. During the manu-
facturing process, PVP polymer chains are entangled
in the polymeric network and become an integral part
of the membrane, and extraction of the single com-
pound becomes almost impossible. In particular, the
composition of the membrane at the interface with
blood represents the crucial issue for biocompatibility
[56] and mass transport efficiency.
From the above observations made during the
evolution of this membrane family, it appears evident
that the inner layer of the membrane is probably the
most important structural component, and its surface
governs the interactions with plasmatic and cellular
blood components to a major degree. In particular,
careful attention has been paid to ensuring the
smoothness of the surface and its chemical modifi-
cation, with creation of hydrophilic–hydrophobic
microdomains [51,52] synergizing to balance minimal
activation of blood components.
Direct and indirect methods to characterize
chemical heterogenicity of the surface and the micro-
domain concept have been applied, including selective
staining of hydrophilic compounds for backscatter
scanning electron microscopy, contact angle hysteresis
C. Ronco et al.
Downloaded from https://academic.oup.com/ndt/article/18/suppl_7/vii10/1818267 by guest on 16 April 2024
measurements and atomic force microscopy using
different detection modes. The latter is also used in
low force tapping mode to study the surface rough-
ness and the porosity/pore distribution of the active
separation layer of the membrane. In Figure 4, a
typical atomic force micrograph (tapping mode) of the
inner surface of a Polyflux high-flux membrane (scan
size: 5
5 mm) is demonstrated.
Mathematical integration of the data allows calcula-
tion of the difference between the highest and lowest
spot (Rz) of the area analysed and two average rough-
ness parameters. The following data were obtained
for Polyflux membrane: Ra ¼ 4.9 nm, Rq ¼ 6.3 nm and
Rz ¼ 68 nm. Such roughness analysis has also been
carried out for other membranes and previously
reported in the literature, and it represents an impor-
tant means to describe the smoothness of the interface
with blood [57].
To study the pore size distribution of nanoporous
polymeric membranes, a whole group of analytical
techniques are available. However, depending on the
technique and the assumptions behind this tech-
nique, different sets of data can be obtained. From
the analysis conducted using permporometry,
The Polyflux family of synthetic membranes
specific statistical distribution curves can be obtained
(Figure 5), which finally allow predictions on the
impact on hydraulic permeability and sieving coeffi-
cients.
Sieving data for single components or sieving curves
for mixtures of substances with different molecular
weight allow characterization of the width of the pore,
size distribution or the sieving properties under defined
conditions. Typical sieving curves for the high- and
low-flux Polyflux membrane types are shown in
Figure 6. However, not only pore size distribution is
important for the final membrane performance, but
also the density of pores (number per unit of surface),
pore length and tortuosity.
Of the different factors determining water perme-
ability and high sieving capacity, pore size has the
largest impact. Small increases in pore size have large
effects on water permeability. A high degree of
membrane porosity is characterized by more pores
per unit of surface area and less pore tortuosity.
Fig. 5. Pore size distribution (schematically) for a low- and high-flux
membrane. Hydraulic permeability is governed by the equation
described at the top where: Kf is the hydraulic permeability, n is the
number of pores, r is the pore radius,  is pore tortuosity, m is
viscosity of the filtrate and
x is the membrane thickness (pore linear
length).
Fig. 6. Sieving curve of Polyflux (high- and low-flux) membranes determined using an aqueous polydisperse dextran solution.
vii15
‘Opening-up’ the membrane structure would increase
large solute permeability, but the detoxification process
in haemodialysis is based predominantly upon size
exclusion (i.e. sieving characteristics of the membrane),
and increasing the mean pore size alone is insufficient
and possibly harmful (i.e. leakage and loss of useful
substances from blood into the dialysate). Basically, a
maximal opening of the pores such as to achieve
maximal removal of large molecules would be ideal, but
this should be obtained in conjunction with a sharp cut-
off in the range of molecular weight such as to exclude
albumin leakage.
These aspects have been accomplished in the past
partially as a result of a specific manufacturing
Downloaded from https://academic.oup.com/ndt/article/18/suppl_7/vii10/1818267 by guest on 16 April 2024
condition particularly related to the choice of the
polymer and the composition of the polymeric blend,
process conditions, temperature, drying and steriliza-
tion. Today, a more accurate and precise control of the
innermost layer of the membrane and its porosity and
surface can be achieved thanks to the recently emerging
and technically applied nano-technologies [57–59]. The
application of new technology principles to define,
more specifically, the key (pore-related) parameters
that influence water and solute permeability is made
possible by progress in the understanding and control
of the chemical engineering at the nano-scale level. In
general, the term ‘nano-technology’ defines the usage of
any technology and processes to produce features
reproducibly at the nanometre scale, e.g. thin films, fine
particles, chemical synthesis on chips with high lateral
resolution, advanced micro-lithography, etc. This
allows for molecular manufacturing processes, i.e. the
creation of a surface and structure modification of the
membrane controlling geometry and distribution of the
pores of the innermost surface at the nano-scale level.
This also includes adaptation of the spinning condi-
tions, improved design of spinning nozzles and finally
results in a more homogeneous polymer solution
mixing, refined precipitation conditions, change of
defined pore structure of the thin inner ‘skin’ region,
fibre dimensions (i.e. optimal internal diameter in
vii16
relation to wall thickness) and optimized fibre structure
for enhanced mass transfer, e.g. ‘wavy’ fibres. A further
control is exerted on pore density and tortuosity with
reduction of mass transfer resistance and improvement
in selectivity of transport characteristics.
All these manufacturing processes controlling struc-
tural features on the nano-scale result in fact in a new
generation of membranes with high selectivity of mass
transport characterized by maximal permeability and
minimal albumin losses, and thus actually approaching
a remarkable similarity to the functional properties of
the glomerular filtration step.
Membrane function in clinical practice
The technical and biophysical characteristics of
membranes affect the permeability characteristics
observed in the clinical setting, although the opera-
tional characteristics of the treatment are often the
most important variable. In particular, hydraulic
permeability and sieving properties of membranes are
strongly determined by blood flow conditions and
blood composition. Membranes, as they should work
as barriers, tend to interact with proteins. However,
when low blood flows are present, the protein
interaction at the blood–membrane interface tends to
be greater. This is due mainly to a lower velocity
gradient between blood and the fibre wall (wall shear
rate). This lower shear effect will result in thicker
boundary layers of proteins and greater concentration
polarization. This effect will in turn result in lower
hydraulic permeability and lower observed sieving
coefficients. Wall shear rates are also affected by blood
viscosity, haematocrit and plasma protein concentra-
tion. There is, for example, a lower flow velocity in
peripheral fibres of a haemodialyser when blood
viscosity is higher. All these problems are enhanced
when high filtration fractions are obtained in hollow
fibre haemodialysers. High filtration fractions result
from increased filtration rates at a given plasma flow.
This is the case when low blood flows are present in the
system and convective therapies are prescribed with
excessive ultrafiltration rates. This unphysiological
condition is deleterious for the hollow fibre, which
becomes progressively clogged and displays a reduction
in permeability and sieving coefficients. Considering
this, convective therapies should be performed only
when blood flows are sufficiently high or in the case of
large amounts of pre-dilution.
Haemodialysis membranes do not remove solutes by
filtration or diffusion alone. Adsorption is an additional
mechanism that may contribute to the final amount
of solute removal. The effect of adsorption strongly
depends on the nature of the membrane, i.e. materials
used to manufacture the membrane. In synthetic
membranes, not only can solutes interact with the
inner surface of the membrane, but they can be
adsorbed onto the supporting external membrane layer.
This external component probably does not affect
the net solute removal since the adsorbed molecules
C. Ronco et al.
would have gone in the filtrate anyway. This external
membrane structure, however, is responsible for
retaining molecules which cannot be found in the
filtrate, making the calculation of sieving coefficients or
direct dialysate quantification methods worthless in
accounting for solute removal. The typical case occurs
with 2-microglobulin, in which sieving coefficient
measurements are extremely fallacious. Synthetic
membranes differ from each other with respect to
protein adsorption. New membrane generations pre-
pared from hydrophobic/hydrophilic electrically neu-
tral polymers show very low values for protein
adsorption [60,61].
However, adsorption appears to become an impor-
Downloaded from https://academic.oup.com/ndt/article/18/suppl_7/vii10/1818267 by guest on 16 April 2024
tant issue for removing endotoxin and other bacteria-
derived products from the dialysate compartment
during fluid filtration. This mechanism in fact not
only permits avoidance of the dangerous effects of
backfiltration with contaminated solutions, but it is
also used for repeated filtration of dialysate to obtain
ultrapure fluid for reinfusion/substitution. Based on
this concept, on-line haemofiltration and haemodiafil-
tration today are performed with a remarkable degree
of safety and clinical tolerance. The problem is mainly
that of ensuring a continuous quality control and
redundant efficiency so that any accidental rupture or
membrane leakage can be overcome by a subsequent
stage of dialysate filtration. Further degrees of safety
can be provided by an on-line system using a third step
filtration in a specific ultrafilter validated for endotoxin
and bacterial retention.
The membrane and the haemodialyser design
The membrane cannot be considered as a single entity
and simply as a biomaterial. It is the barrier for solute
separation processes and it may perform very
differently depending on the device in which it is
employed [62–70].
For this reason, the membrane should be seen as a
component in a more complex therapeutic system
which includes the blood compartment and the
dialysate compartment.
The blood compartment of an ideal haemodialyser
should be characterized by the compromise between a
low priming volume and a maximal blood–membrane
contact surface [60]. The hollow fibre design has
certainly contributed to achieving better results
compared with the flat sheet configuration. The blood
ports, considered for many years simply as arterial and
venous ends of the unit, today represent sites of great
importance. The arterial port must have a minimal
stagnation of flow and it should guarantee a homo-
geneous distribution of blood flow in all the fibres of the
bundle. For this purpose, different types of flow
distributors have been proposed (conical, spiral, etc.)
with reduced space between the cap and the potting.
No dead spaces or irregularities in the internal surface
should be present. All these features should be accom-
panied by an accurate cutting of the fibres to (i) prevent
The Polyflux family of synthetic membranes
any collapse or accidental obstruction of the fibre
internal lumen and (ii) create a smooth surface to
reduce cell activation. The sterilization procedure may
be critical in preserving the structure of the polyure-
thane utilized for the potting.
The length and the number of fibres in a dialyser
characterize the hydraulic resistance of the filter at a
given blood flow. In treatments utilizing high rates of
convection, filtration pressure equilibrium may occur
along the length of the filter. As blood moves through
the filter, water is removed by filtration, and haemato-
crit and plasma protein tend to increase. As a conse-
quence, the oncotic power/pressure of plasma proteins
becomes as high as the internal hydrostatic pressure,
and filtration ceases. This phenomenon causes possible
disturbances/inconvenience in the distal part of the
filter where a highly viscous blood is flowing inside the
hollow fibres. An accelerated rate of clotting or
alteration in flow distribution may occur under such
circumstances. Today, with the increased average
haematocrit up to 35% due to the broad and
common use of erythropoietin to correct anaemia, all
dialysers are forced to operate under conditions of
highly viscous blood. As a result, the entire system may
be affected and it may be likely that peripheral fibres
operate with much lower flows compared with the
central fibres of the bundle. This results in lower
velocity gradients at the blood–membrane interface
(low wall shear rates) and less than optimal operational
conditions of those fibres, including prolonged contact
time, which is thought specifically to cause clot
propagation.
To adapt the hydraulic resistance of the haemodia-
lysers to the different operational conditions and the
different treatment techniques, hollow fibres with
different inner diameters have been designed. In an
attempt to reduce the blood compartment resistance,
the inner diameter has been increased in some filters
from 200 to 250 mm [63]. These filters have been used
mostly in arterio-venous circuits, as well as in patients
treated with continuous renal replacement therapies. In
filters for babies or neonates, the inner diameter has
been increased up to 500 or even 1000 mm [64]. The
reason for these approaches has been to attempt to
reduce the rate of obligatory filtration at a given blood
flow. In other words, because of a lower resistance,
these filters permit higher blood flows at a given arterio-
venous pressure gradient (in arterio-venous circuits)
and they present a lower internal pressure drop in
the presence of a pumped extracorporeal circuit with
a consequent lower rate of obligatory filtration.
Obligatory filtration depends on the pressure generated
inside the filter: at a constant venous pressure, the
higher the resistance of the filter, the higher will be
the pressure drop in the filter and the higher will be the
pressure at the inlet of the filter. As a consequence, the
average pressure in the blood compartment will be
higher and so will the filtration rate at a given blood
flow and venous pressure [65,66].
The opposite approach has been suggested in an
attempt to maintain very high wall shear rates and
vii17
to increase the obligatory filtration. This has been
obtained reducing the internal diameter down to
175–185 mm [67]. In these circumstances, at a similar
number of fibres, the cross-sectional area of the filter
will be reduced and the average flow velocity per fibre
will increase. At the same time, the pressure drop in the
blood compartment will significantly increase. With the
use of dialysis machines with accurate ultrafiltration
control systems, this will result in higher rates of
filtration–backfiltration at a given net filtration rate.
This will increase the convective component of the
transport, without requiring replacement fluid reinfu-
sion as in the case of haemodiafiltration. A similar
approach has been proposed recently by placing a
Downloaded from https://academic.oup.com/ndt/article/18/suppl_7/vii10/1818267 by guest on 16 April 2024
constriction ring around the fibre bundle in the mid-
segment of the dialyser, with the aim of increasing the
pressure drop both in the blood and in the dialysate
compartment [65] (technically not applied yet in regular
dialysis). However, from the bio-engineering point of
view, it is obvious that membrane configuration and
device design can be fine-tuned to optimize perfor-
mance further.
The design and the construction of the dialysate
compartment is of major importance for the perfor-
mance of a dialyser. Several attempts have been made
in the past to improve the housing design (including
dialysate distribution) and the fibre distribution to
optimize mass transport. Some modifications have been
introduced over the years such as the position of the
Hansen connectors in the case (same side or opposite
side). Initial studies on solute clearances at different
dialysate flows demonstrated that by increasing
dialysate flow over a certain limit (600 ml/min) with
standard dialysers, the clearance of small molecules
could be negatively affected [9]. This effect was
attributed to a channelling phenomenon that was
creating a preferential flow of dialysate in the region
external to the bundle with consequent stagnation in
the internal region of the haemodialyser. The first
attempt to reduce this phenomenon was the application
of an ‘overflow’ ring internal to the case, to force the
dialysate flow into the internal regions of the bundle.
Other types of distributors were designed for this
purpose, achieving an improved performance of the
dialyser at higher dialysate flow rates. More recently,
studies on flow distribution have pointed out the
importance of the bundle design per se [68–71]. First of
all, the density of the fibres inside the case affects the
resistance to dialysate flow and may contribute to a
variable distribution of dialysate flow within the fibres.
Furthermore, the phenomenon of ‘packing’ of the
fibres is responsible for an inhomogeneous distribution
of the flow and possible stagnation in various segments
of the bundle. The final effect of stagnation may be the
reduction of the gradients for diffusion. However,
another condition in which the gradient for diffusion is
reduced is when high convective rates are utilized in
haemodiafiltration. Under such circumstances in fact,
ultrafiltrate is transported on the other side of the
membrane with a solute concentration similar to that
of plasma water. This will result in similar solute
vii18
Fig. 7. Cross-section through the header area of the Polyflux dialyser with the undulated and crossed fibres.
concentrations on both sides of the membrane, and
diffusion ceases. The phenomenon is aggravated
further if fresh dialysis solution is not flowing
homogeneously on the dialysate side external to the
fibres washing away the ultrafiltrate. In these condi-
tions, stagnation will occur and the performance of the
haemodialyser will be negatively affected.
To prevent this phenomenon, at least three
techniques have been applied [68,71]. The first consists
of the placement of space yarns between the fibres to
create a better distribution path of the dialysate
solution. In the second technique [71], a special fibre
crossing with a certain angle is applied during bundle
formation to obtain even dialysate distributions. The
third method is the creation of the so-called ‘Moiré
structure’ which consists of a waved shape (undulation)
of the hollow fibres that prevents the packing of the
fibres and maintains the dialysate path open in the
entire cross-sectional area of the bundle. In the Polyflux
family, waved fibres are additionally crossed and, by
combining these two steps, a three-dimensional net-
work guarantees stable and high mass transport rates in
the whole bundle (see Figure 7). Not only is a
homogeneous distribution of the flow obtained with
these approaches, but significant increases in solute
clearances can also be achieved.
In conclusion, the membrane structure and material
certainly affect the performance of the haemodialysis
process. When high convective rates are prescribed,
high-flux synthetic membranes must be utilized. Their
performances are strongly affected by several factors
other than the original material, such as geometry of
the haemodialyser and hollow fibre design. Thanks to
the improved design and membrane characteristics,
newer treatments such as on-line haemofiltration and
haemodiafiltration can be easily performed. Higher
C. Ronco et al.
Downloaded from https://academic.oup.com/ndt/article/18/suppl_7/vii10/1818267 by guest on 16 April 2024
convection rates will permit us to understand the
effective benefit of convection over diffusion, without
facing the limitations imposed in the past by the
inadequate technology available.
References
1. Singer SJ, Nicolson GL. The fluid mosaic model of the
structure of cell membranes. Science 1972; 175: 720–731
2. Kolff WJ. Dialysis in the treatment of uremia. Arch Intern Med
1954; 94: 142–147
3. Bywaters EGL, Joeckes BM. The artificial kidney: its clinical
application in the treatment of traumatic anuria. Proc R Soc
Med 1948; 41: 411–420
4. Scribner BH, Buri R, Caner JEZ, Hegstrom R, Burnell JM.
The treatment of chronic uremia by means of intermittent
hemodialysis: a preliminary report. Trans Am Soc Artif Intern
Org 1960; 6: 114–120
5. Quinton WE, Dillard D, Scribner BH. Cannulation of blood
vessels for prolonged hemodialysis. Trans Am Soc Artif Intern
Org 1960; 6: 104–108
6. Kiil F, Amundsen B. Development of a parallel flow artificial
kidney in plastics. Acta Chir Scand 1960; 253 [Suppl]: 142–149
7. Scribner Bh, Caner JEZ, Buri R, Quinton WE. The technique
of continuous hemodialysis. Trans Am Soc Artif Intern Org
1960; 6: 88–94
8. Ramirez O, Swartz C, Onesti G, Mailloux L, Brest AN. The
winged in-line shunt. Trans Am Soc Artif Intern Org 1966; 12:
220–223
9. Brescia MJ, Cimino JE, Appel K, Hurwick BJ. Chronic
hemodialysis using venapuncture and a surgically created
arterio-venous fistula. N Engl J Med 1966; 275: 1089–1091
10. Mion CM, Hegstrom RM, Boen ST, Scribner BH. Substitution
of sodium acetate for bicarbonate in the bath fluid for
hemodialysis. Trans Am Soc Artif Intern Org 1964; 10: 110–114
11. Scribner BH, Babb AL. Chronic hemodialysis in Seattle
1960–1966, part II. Dial Transplant 1982; 11: 324–330
12. Merrill JP, Schupak E, Cameron E, Hampers CL. Hemodialysis
in the home. Trans Am Soc Artif Intern Org 1965; 11: 7–12
The Polyflux family of synthetic membranes
13. Babb AL, Popovich RP, Christopher TG, Scribner BH. The
genesis of the square meter-hour hypothesis. Trans Am Soc
Artif Intern Org 1971; 17: 81–86
14. Babb AL, Farrell PC, Uvelli DA, Scribner BH. Hemodialyzer
evaluation by solute molecular spectra. Trans Am Soc Artif
Intern Org 1972; 18: 98–103
15. Funck-Brentano JL. Experience with a new open membrane.
Proc Clin Dial Transplant Forum 1971; 1: 56–63
16. Gurland HJ, Brunner FP, Chantkler C et al. Combined report
on regular dialysis and transplantation in Europe VI. Proc Eur
Dial Transplant Assoc 1976; 13: 3–9
17. Cambi V, Savazzi G, Arisi L et al. Short dialysis schedule
(SDS)—finally ready to become routine? Proc Eur Dial
Transplant Assoc 1974; 11: 112–118
18. Reiger J, Quellhorst E, Lowitz HD, Kong RG, Scheler F.
Ultrafiltration for middle molecules in uremia. Proc Eur Dial
Transplant Assoc 1974; 11: 158–165
19. Henderson LW, Colton CK, Ford CA. Kinetics of hemodiafil-
tration: clinical characterization of a new blood cleansing
modality. J Lab Clin Med 1975; 85: 355–361
20. Babb AL, Strand MJ, Uvelli DA, Multinovic J, Scribner BH.
Quantitative description of dialysis treatment: a dialysis index.
Kidney Int 1975; 7 [Suppl 2]: S23–S27
21. Scribner BH. Substitution of bicarbonate for acetate in the
dialysate for the care of a critically ill patient. Dial Transplant
1977; 6: 26–30
22. Ward RA, Wathen RL, Williams TE. Effects of long-term
bicarbonate hemodialysis (BHD) on acid–base status. Trans
Am Soc Artif Intern Org 1982; 28: 295–301
23. Gotch F, Sargent JA. A mechanistic analysis of the National
Cooperative Dialysis Study (NCDS). Kidney Int 1985; 28:
526–531
24. Rotellar E, Martinez E, Samsò JM et al. Why dialyze more
than six hours a week? Trans Am Soc Artif Intern Org 1985; 31:
538–541
25. Miller JH, von Albertini B, Gardner PW, Shinaberger JH.
Technical aspects of high flux hemodiafiltration for adequate
short (under 2 hours) treatment. Trans Am Soc Artif Intern Org
1984; 30: 377–380
26. Keshaviah P, Collins A. Rapid high efficiency hemodialysis.
Trans Am Soc Artif Intern Org 1986; 32: 17–21
27. Ronco C, Fabris A, Chiaramonte S et al. Comparison of four
different short hemodialysis techniques. Int J Artif Org 1988; 3:
169–174
28. Ronco C, Fabris A, Chiaramonte S et al. Impact of high blood
flows on vascular stability in hemodialysis. Nephrol Dial
Transplant 1990; 5 [Suppl 5]: 109–114
29. Tolchin DO. Acetate metabolism and high efficiency hemodia-
lysis. Int J Artif Org 1979; 2: 1–6
30. Ronco C, Fabris A, Feriani M et al. Technical and clinical
evaluation of a new system for ultrafiltration control during
hemodialysis. Trans Am Soc Artif Intern Org 1988; 3: 613–615
31. Goehl H, Konstantin H, Gulberg CA. Hemofiltration
membranes. Contrib Nephrol 1982; 32: 20–30,
32. Bosch JP, von Albertini B, Ronco C. Hemofiltration and hemo-
diafiltration. In: Glassock RJ, ed. Current Therapy in Nephrology
and Hypertension 2. BC Decker Inc., Toronto; 1987; 258–263
33. Floege J, Granolleras C, Smeby L, Goehl H, Koch KM,
Shaldon S. Hydrophilic high-flux polyamide membranes are
more efficient for beta-2-microglobulin (b2m) removal than
hydrophobic polyamide membranes. Abstracts ASAIO 1987; 40
34. Goehl H, Buck R, Strathmann H. Basic features of the
polyamide membranes. Contrib Nephrol 1992; 96: 1–25
35. Sternby J, Joensson S, Ledebo I. Hemodiafiltration: technical
aspects. Contrib Nephrol 1992; 96: 86–98
36. Ledebo I. On-line hemofiltration: technique and therapy. Adv
Renal Replacement Ther 1999; 6: 195–208
37. Santoro A, Spongano M, Ferrari G, Zucchelli P. Analysis of
the factors influencing bicarbonate balance during acetate free
biofiltration. Kidney Int 1993; 43 [Suppl 41]: S184–S187
vii19
38. Ghezzi PM, Frigato G, Fantini GF et al. Theoretical model
and first clinical results of the Paired Filtration Dialysis. Life
Support Syst 1983; [Suppl. 1]: 271–278
39. Feriani M, Ronco C, Biasioli S, Bragantini L, La Greca G.
Effect of dialysate and substitution fluid buffer on buffer flux in
hemodiafiltration. Kidney Int 1991; 39: 711–717
40. Canaud B, Imbert E, Kaaki M et al. Clinical and microbiological
evaluation of a postdilutional hemofiltration system with inline
production of substitution fluid. Blood Purif 1990; 8: 160–170
41. Ledebo I. Development to supply endotoxin-free dialysate and
substitution fluid. Artif Org 1993; 17: 397
42. Shaldon S, Beau MC, Branger B et al. Economic preparation
of sterile non-pyrogenic infusate for hemofiltration. Dial
Transplant 1983; 12: 792–793
43. Ronco C. Backfiltration in clinical dialysis: nature of the
phenomenon, mechanisms and possible solutions. Int J Artif
Downloaded from https://academic.oup.com/ndt/article/18/suppl_7/vii10/1818267 by guest on 16 April 2024
Org 1990; 13: 11–21
44. Deppisch R, Haug U, Mientus W, Goehl H. Contact-phase
activation in dialysis: relationship between membrane surface
properties and activation mechanism. Blood Purif 1992; 10: 87
45. Deppisch R, Ritz E, Hänsch GM, Schöls M, Rauterberg EW.
Bioincompatibility—perspectives in 1993. Kidney Int 1994; 55:
S77–S84
46. Ronco C, Brendolan A, Crepaldi C. Water and dialysate
treatment for hemodialysis. In: Bosch JP, ed. Hemodialysis;
High Efficiency Treatments. Churchill Livingstone, New York;
1993: 41–61
47. Schindler R, Lufft V, Schulze M. Hemodialysis using
regenerated cellulose but not polyamide induces IL-1 mRNA
in mononuclear cells in vivo. Blood Purif 1991; 9: 46
48. Strathmann H, Goehl H. Membranes for blood purification:
state of the art and new developments. Contrib Nephrol 1990;
78: 119–141
49. Goehl H, Storr M, Deppisch R. New generation of steam
sterilized dialysers. Blood Purif 1998; 16: 235
50. Storr M, Deppisch R, Buck R, Goehl H. The evolution of
membranes for hemodialysis. In: Hincal AA, Kas HS, eds.
Biomedical Science and Technology. Plenum Press, New York;
1998: 219–233
51. Deppisch R, Buck R, Dietrich R, Storr M, Goehl H, Smeby L.
Microdomain-structured synthetic polymer surfaces: applica-
tion, performance and potential. Int J Artif Org 1997; 20: 533
52. Hoenich NA, Stamp S, Robert S. Clinical characterisation of a
micro domain structured synthetic membrane for haemodia-
lysis. Artif Org 1999; 23: 650
53. Ward RA, Schmidt B, Samtleben W. Alloying with a heat
resistant polymer allows steam sterilization and improves the
performance of polyamide membranes [abstract]. J Am Soc
Nephrol 1998; 9: 186A
54. Lonnemann G, Behme TC, Lenzner B et al. Permeability of
dialyzer membranes to TNFalpha-inducing substances derived
from water bacteria. Kidney Int 1992; 42: 61–68
55. Weber C, Linsberger I, Rafiee-Tehrani M, Falkenhagen D.
Permeability and adsorption capacity of dialysis membranes to
lipid A. Int J Artif Org 1997; 20: 144–152
56. Berger ED, Kaiser J, Schaefer K, Risler T. Biocompatibility
evaluation of a new synthetic steam sterilized low-flux dialyzer
[abstract]. J Am Soc Nephrol 1998; 9: 167A
57. Bowry SK, Ronco C. Surface topography and surface elemental
composition analysis of HelixoneÕ, a new high-flux polysulfone
dialysis membrane. Int J Artif Org 2001; 24: 757–764
58. Ronco C, Bowry S. Nanoscale modulation of the pore
dimensions, size distribution and structure of a new
polysulfone-based high-flux dialysis membrane. Int J Artif Org
2001; 24: 726–735
59. Ronco C, Nissenson AR. Does nanotechnology apply to
dialysis? Blood Purif 2001; 19: 347–352
60. Chanard J, Caudwell V, Lavaud S, Hanrotel C, Canivet E,
Vincent C. Assessment of 2-microglobulin ( 2m) binding onto
dialysis membranes using 131I- 2m. J Am Soc Nephrol 1994; 5: 438
vii20
61. Goldman M, Lagmiche M, Dhaene M, Amraoui Z,
Thayse C, Vanherweghem JL. Adsorption of 2-microglobulin
on dialysis membranes: comparison of different dialyzers
and effects of reuse procedures. Int J Artif Org 1989; 12:
373–378
62. Ronco C, Ghezzi P.M, Brendolan A, Crepaldi C, La Greca G.
The haemodialysis system: basic mechanism of water and solute
transport in extracorporeal renal replacement therapies.
Nephrol Dial Transplant 1998; 13 [Suppl 6]: 3–9
63. Ronco C, Brendolan A, Crepaldi C. Importance of hollow fiber
geometry in continuous arterio-venous hemofiltration. Contrib
Nephrol 1991; 91: 175–178
64. Ronco C, Parenzan L. Acute renal failure in infancy: treatment
by continuous renal replacement therapy. Intensive Care Med
1995; 21: 490–499
65. Ronco C, Orlandini G, Brendolan A, Lupi A, La Greca G.
Enhancement of convective transport by internal filtration in a
modified experimental hemodialyzer. Kidney Int 1998; 54:
979–985
C. Ronco et al.
66. Brendolan A, Ronco C, Ghezzi PM, Scabardi M, La Greca G.
Hydraulic and flow-dynamic characteristics of PMMA dialyz-
ers. Contrib Nephrol 1998; 125: 41–52
67. Ronco C, Brendolan A, Lupi A, Metry G. Effects of a reduced
inner diameter of hollow fibers in hemodialyzers. Kidney Int
2000; 58: 809–817
68. Ronco C, Scabardi M, Goldoni M, Brendolan A, Crepaldi C,
La Greca G. Impact of spacing filaments external to hollow
fibers on dialysate flow distribution and dialyzer performance.
Int J Artif Org 1997; 20: 261–266
69. Ronco C, Brendolan A, Crepaldi C, Rodighiero M, Scabardi M.
Blood and dialysate flow distributions in hollow-fiber hemo-
dialyzers analyzed by computerized helical scanning technique.
J Am Soc Nephrol 2002; 13: S53–S61
70. Ronco C, Bowry SK, Brendolan A et al. Hemodialyzer: from
macrodesign to membrane nanostructure; the case of the FX-
Downloaded from https://academic.oup.com/ndt/article/18/suppl_7/vii10/1818267 by guest on 16 April 2024
class of hemodialyzers.
71. Goehl H, Raff M, Harttig H, Deppisch R. PC-PE hollow-fiber
membrane. Blood Purif 1986; 4: 23–31