вход по аккаунту



код для вставкиСкачать
On-Line Hemodialiltration
Maeda K, Shinzato T (eds): Effective Hemodiafiltration: New Methods.
Contrib Nephcol. Basel, Karger, 1994, vol 108, pp 1-11
A Decade of Experience with On-Line
Gambro AB, Lund, Sweden
Convective transport across dialysis membranes has been known for a long
time to be a good alternative to diffusion. The interest in hemofiltration (HF) as a
blood cleaning procedure has thus persisted in spite of practical difficulties. In
fact, the term 'diafiltration' was used already in 1967 by Henderson et al. [1], but
as a name for a procedure that we would call predilution HF today.
Several benefits ofHF over normal hemodialysis (HD) have been well documented. Some of these are: better cardiovascular stability [2], higher survival rate
of high risk patients [3, 4] and more removal of high molecular weight substances
(such as /32-microgiobulin, /3z-m) [5] .
With better dialysis techniques, such as the increased use of bicarbonate
buffer and machines with ultraftltration (UF) volume control, some of the advantages of HF over normal HD are not so evident any more. Today, mainly due to
the discovery [6] of /32-m amyloidosis, increased emphasis has been based on the
removal oflarge molecules. Here, the use of convection has clear advantages, and
the problem is how to reach sufficient exchange volumes. This has in fact always
been the main issue in conventional HF, where the achievable UF rates have
limited the treatment efficiency (i.e. the urea and creatinine clearances).
There are several factors that limit what can be achieved by conventional HF
as it is usually performed today. Examples are blood flow, hematocrit, filter performance, buffer and costs. It should also be said that exactly the same limitations
Downloaded by:
Vanderbilt University Library - 10/27/2017 1:59:38 PM
Limitations with Conventional Hemofiltration
are valid for the removal of large molecules by hemodiafiltration (HDF). Most
important are the factors that limit the achievable UF rate, since this will determine clearance.
Blood flow is an obvious factor, the main one, just as with normal HD. But in
HF this factor is even more limiting. The problem is that it is impossible to concentrate the blood more than to a certain hematocrit level. This was shown theoretically and experimentally for some specific setups [7], with a limit of 50-60%
depending on membrane, blood composition and blood flow. Without a high
blood flow it is thus impossible to get a high UF rate.
With a given hematocrit of untreated blood, the limit above imposes a maximum on the obtainable filtration factor (which can be used to estimate the achievable clearance). Therefore, the hematocrit of the incoming blood also comes in as
an important limitation. With the increased use of erythropoietin this factor has
become even more important. What happens is that together with the proteins,
the blood cells will form a secondary layer. This will decrease the membrane permeability, and it also shows up as an increased pressure drop along the dialyzer
[8]. Altogether this means that it will be difficult to get high enough UF rate and
urea clearance.
The buffer used in commercially available substitution fluids is usually not
bicarbonate. This is because of problems with calcium precipitation. Instead lactate or acetate are used, but they have to be metabolized in the body to produce
bicarbonate. The slow metabolization of acetate, and partly also lactate, limits the
maximum infusion rate that can be allowed without detrimental effects to the
The costs of preprepared substitution fluids are also a problem. This, and the
practical problems of handling say 50 liters of fluid in bags or bottles, has often
limited the exchange volumes used, both in HDF and HF.
One solution to the problem oflow urea clearance in conventional HF is to
add diffusive transport by combining filtration and dialysis. The clearance
obtained then is shown in figure 1 (left part) as a function ofUF rate. The curves
are purely theoretical, but based on data for real high-flux dialyzers [8]. The upper
curve may represent a small molecule such as urea, and the lower curve is for
substances with a molecular weight like inulin.
First of all, we may note that the relative increase in clearance with UF is
much larger for large solutes. This is natural, since their diffusive clearance is low.
But it is also obvious that quite a high UF rate is needed in order to get any
substantial improvement in urea clearance. This is difficult to accomplish with
Downloaded by:
Vanderbilt University Library - 10/27/2017 1:59:38 PM
HDF clearance
o - Pull
x - Push
* - Mean
Push-pull clearance
UF rate (ml/min)
Large solute
UF rate (ml/min)
standard HDF. An infusion volume of e.g. 9.6 liters over 4 h gives only 40 ml!
min, and a fairly small increase in clearance. For larger molecules, the relative
increase in clearance is larger, but still insufficient to reduce e.g. ~2-m to acceptable levels, and only a small fraction of what could be achieved if more substitution
fluid were available.
Another alternative, practised in Japan, is push/pull HDF [9]. Figure I (right
part) shows the theoretical clearance for this form of treatment, again for the same
small and large solutes. The dashed curves are the clearances for normal HDF, but
now extended to cover also negative UF rates. This is needed, since we must have
a high negative UF rate during the push phase to compensate for the high positive
UF rate during the pull phase.
For small solutes, the difference between normal HD and push/pull treatment is negligible. But for larger solutes, there is a substantial improvement. In
the case shown in figure 1, clearance is about twice as high for push/pull HDF as
for normal HD. We can also see, however, that with the same high UF used in the
pull phase of push/pull HDF, standard HDF is always more effective, both for
small and large solutes. It corresponds to staying in the pull phase for the whole
It can be concluded that although shifting from conventional HF to HDF can
improve the clearance of small molecules like urea, it does not help for bigger
molecules. The same large exchange volumes are needed in both HF and HDF in
order to get a good clearance of such substances.
A Decade of Experience with On-Line HF/HDF
Downloaded by:
Vanderbilt University Library - 10/27/2017 1:59:38 PM
Fig. 1. Theoretical clearances as a function ofUF in standard HDF (left) and in push/
pull HDF [9] (right). The total clearance in push/pull HDF is found along the straight line
connecting high UF (0) and the backfiltration UF (x), at the UF rate corresponding to the
desired weight loss (*), here chosen to be zero.
New Hemofiltration
In adding dialysis fluid to the dialyzer, converting the treatment into HDF,
we may lose some of the benefits of HF. Haas et al. [10] compared HDF and
predilution (PRD) HF. In spite of a slightly higher weight loss, the incidence of
hypotension was much lower with HF. There was no difference in blood flow,
buffer, filter or treatment time.
We would therefore like to keep the benefits ofHF, but remove the drawback
oflow urea clearance. Today, conventional HF treatments are performed in postdilution (POD) mode. The main reason is to save substitution fluid. Clearance is
then roughly equal to UF rate, at least for substances with a sieving coefficient of
1; and since UF is limited to 25-40% of the blood flow, this also gives a limit to
the achievable urea clearance.
Now if we turn to pre dilution HF instead, clearance for substances with a
sieving coefficient of 1 is given by the formula
Clearance = QUF * Dilution factor = QUF *
Downloaded by:
Vanderbilt University Library - 10/27/2017 1:59:38 PM
where QB and QINF are the blood flow and infusion flow. QUF is the total UF
across the membrane, which equals the sum of QINF and fluid removal from the
patient. To understand the formula, one must remember that POD clearance is
given by the ultrafiltration QUF. Here QUF is multiplied by the dilution factor for
blood at the dialyzer inlet, since the concentrations in the ultrafiltrate will be
multiplied by this factor. We may also note from experience that infusion rates in
the range of the blood flow can usually be reached. Then the factor QUFI
(QB+QINF) is around 0.5, and clearance will be about 50% of the blood flow, and
sometimes even higher. This is a significant improvement over POD.
The clinical examples shown in figure 2 illustrate the differences between
POD and PRD treatments that have been performed under similar conditions.
The blood flow was 400 mUmin in both cases, and the same type of filter was used.
Transmembrane pressure (TMP) was 250 mm Hg in POD and 200 mm Hg in
PRD modes. In POD, the achieved UF was only 100 ml/min. Clearance is thus also
100 ml/min for substances with a sieving coefficient of I. In PRD however, UF was
equal to blood flow, 400 mUmin, giving a clearance of200 mUmin. The difference
between the two modes would have been slightly smaller if a higher TMP had been
used. But the fact remains that it is possible to get a higher clearance with PRD.
Note also the gradual UF decrease in the POD treatment, which is more sensitive
than PRD to this effect of hemoconcentration due to the weight loss.
To explain the high clearances in PRD mode, we can look at what happens to
hematocrit and proteins in the passage through the dialyzer, during which the
blood gets increasingly concentrated. In POD, it reaches its maximum at the
Po tdilutibn HF
j~[ilt: I
!2 I
Predjlution HF
,: . j
Ultrafi ltration
Clearance (S= I)
Clearance (S= I)
,, 200
II ,. I I I I I 1
Treatment time (h)
~ 100
Treatment time (h)
Fig. 2. UF and urea clearance in a POD-HF treatment (left) and in a similar PRD
treatment (right).
A Decade of Experience with On-Line HF/HDF
Downloaded by:
Vanderbilt University Library - 10/27/2017 1:59:38 PM
end of the dialyzer, where it is again diluted. In PRD, however, we start by diluting the blood, maybe in a 1: 1 relation. This reduces the hematocrit at the dialyzer
inlet by 50%. The blood is then gradually concentrated again, and reaches the
original concentration at the dialyzer end. Obviously, the mean hematocrit is
much higher in POD than in PRD. This concentration difference is even more
pronounced for plasma proteins, since the relative changes in plasma volume are
larger than in blood volume. In addition, the effects of concentration on blood
viscosity and UF rate are highly nonlinear.
This leads to a UF rate and clearance that is lower for POD than for PRD.
Figure 3 shows UF as a function of TMP for a polyamide membrane with the
POD case to the left. The diagrams to the right are for blood that is diluted in a 1: 1
relation, which corresponds to our experience with PRD. Note that the scales are
adjusted to be equal for urea clearance (S= I), which in PRD is only half the UF
rate. Thus even though the substitution fluid is used less efficiently in PRD mode,
the UF rate is increased so much that clearance is higher than in POD. In order to
get this effect, the membrane used must have a permeability that is much higher
for water than for blood.
We can also compare achievable clearancesat maximum TMP for POD and
PRD as a function of blood flow, shown schematically in figure 4. For low blood
flows the PRD clearance will approach the blood flow. This is because the filter
capacity will be high enough to allow high dilution factors. POD, however, is even
at low blood flows limited by the high hematocrit at the end of the dialyzer. At
high blood flows PRD will be limited by filter capacity and the advantage ofPRD
may decrease. Where on the curve this happens depends on the blood and the
membrane, but we have little or no experience at these very high blood flows .
undiluted blood (32/60)
UF rate
diluted blood 1:1
UF rate
400+400 mllmin
300+300 mllmin
250 250
200 200
400 mllmin
300 mllmin
3-point TMP (mmHg)
Fig. 3. Measured UF rate and calculated urea clearance vs. TMP for blood flows of 300
and 400 mVmin for undiluted blood (hematocrit = 32% and total protein = 60 gil) and for
blood diluted with an equal amount of Nael (0.9%). A 2-m 2 polyamide filter (FH88H) was
used in both cases.
To summarize, we can state that PRD HF may offer the following possibilities. We can get a high solute removal for large molecules. At the same time,
removal of small solutes will be sufficient, at least with blood flows around 300
mVmin. With the use of bicarbonate buffer in the substitution fluid we can get a
good correction of fluid, electrolyte and acid-base balances. It is also easy to calculate clearance. In fact, by monitoring infusion rate, weight loss and blood flow (as
most machines do), urea clearance, and thereby 'treatment adequacy', can be continuously monitored without additional measurements. HF as such is considered
a more physiological treatment. This has been documented e.g. by better cardiovascular stability. Whether this is valid also in PRD remains to be shown.
On-Line System
Downloaded by:
Vanderbilt University Library - 10/27/2017 1:59:38 PM
Let us now go back to the limiting factors of conventional HF. As mentioned
above, PRD is less limited than POD by the factors related to clearance (blood
flow, hematocrit and filter performance). But PRD requires large quantities of
substitution fluid in order to be effective. This leads to a demand for simple and
cost-effective on-line preparation of fluid. Additionally, the two other problems
with conventional HF can be taken care of at the same time. One can use bicarbonate as a buffer and produce large amounts of fluid at low cost.
.. ' .'
AS (ml/min)
For these reasons we have been developing and refining the on-line concept
for more than a decade. The basic principles are shown in figure 5, where a main
idea is that we should always keep the fluid lines of the machine as clean as possible. We also want some extra safety for the infusion fluid. Both goals are obtained
by multiple filtration/adsorption and effective sterilization. The incoming reverse-osmosis-treated water is filtered before it enters the machine. The final fluid
is then filtered again before it leaves the machine. These polyamide filters
(U7000) are disinfected with the machine between treatments. After the second
U7000 filter the fluid should be safe for infusion, but for added safety there is a
smaller disposable polyamide ultrafilter in the infusion line [11] . It should further
be noted that ultrapure dialysis fluid for normal HD is also achieved with this
system. It has been suggested that the incidence of carpal tunnel syndrome is
lower when ultrapure dialysis fluid is used [12].
The fluid produced may be used in a number of ways. The solid line (fig. 5)
shows a PRD-HF treatment. POD is obtained simply by infusing into the venous
blood line instead of the arterial line. As much of the fluid as needed is pumped
through the disposable filter and infused into the blood line. The unused portion
of the dialysis fluid is collected together with the ultrafiltrate and returned to
drain. HDF results if we allow the rest of the dialysis fluid to pass the dialyzer
before it goes to drain. This can be done in either PRD or POD mode.
The infusion capacity ofthis system is 450 mllmin. The disposable filter in the
infusion line is the limiting factor, but with this infusion flow we will reach a volume
of 100 liters in less than 4 h .This should be more than enough in most cases.
A Decade of Experience with On-Line HF/HDF
Downloaded by:
Vanderbilt University Library - 10/27/2017 1:59:38 PM
Fig. 4. Schematic diagram of clearance vs. blood flow for PRO- and POD-HF.
A concentrate
Fig. 5. Flow diagram of the Gambro AK I00 ULTRA system for preparation of substitution fluid online. RO = Reverse osmosis.
Downloaded by:
Vanderbilt University Library - 10/27/2017 1:59:38 PM
There are some requirements for the continuous production of such a clean
dialysis fluid. We must have a high quality water entering the machine, i.e. water
for dialysis treated by reverse osmosis. Secondly, we must prevent any bacteria
from entering through the concentrates. Liquid bicarbonate-concentrates are
potential sources of contamination. Therefore we use powder bicarbonate dissolved on-line during treatment [13]. The multiple UF steps provide an ultrapure
fluid, provided that the machine and filters are regularly disinfected. The filtration is based on membrane cutoff as well as adsorption to the filter surface. Therefore, the two large filters need to be exchanged at least once a month to guarantee
that there will be no breakthrough.
When not in use, e.g. nights and weekends, the machine is filled with disinfectant; presently we use peracetic acid. Rinsing takes about 40 min , but this does
not bother the user since the machine is programmed to be ready for use at a
certain time each day. This time can be different for different days of the week.It
is also important to note that filling with disinfectant does not guarantee sterility,
unless the system is carefully designed. The choice of materials used is crucial, the
surface structures have to be suitable, and there must be no dead ends in the
machine. The operator should take care in handling machine and disposables and
keep the machine clean. Numerous reports have shown that, if properly handled,
this system is capable of producing fluid that is safe for infusion [14-18].
A substantial amount of experience with on-line fluid production has been
collected over the years. More than 50,000 treatments have been performed with
the different versions of the system, of which more than 2,000 have been documented in clinical studies and/or publications. The first commercial system
(GHS-IO) was used in Berlin [17], Hannover, Montpellier and Parma, and the
second version (MPS-10) in Pisa [19] and Stockholm (internal report). The most
recent version (ULTRA) is based on the AK1 00 system and has now been used in
several clinics for more than 2 years. The total amount of fluid produced by the
different systems from 1981 until the beginning of 1993 adds up to at least 1.3
million liters. Our experience has shown that this is a completely safe way to
produce fluid for infusion.
The efficiency of PRO and POD on-line HF is compared in an on-going
study by Drs. Shaldon and Granolleras in Nimes, and we are grateful to them for
providing these preliminary data. There are two objectives with this study. First
to establish the TMP that will maximize the removal of urea and ~z-m in PROHF. Secondly, this optimized PRD-HF should be compared to POD-HF.
The same patient, a 25-year-old female (10 years on POD-HF) with a Thomas femoral AV shunt, was involved in three treatments for each setting. The
Gambro ULTRA machine and FH88H hemofilter were used in all treatments.
Blood flow was 300 mllmin, hematocrit 30-32% and total protein 73-77 gil and
all studies were performed after the longest interdialytic interval. The restriction
to only 1 patient gives the advantage of having fairly constant blood parameters.
This is important in a comparison like this, since HF results are very dependent
on the blood composition.
Preliminary data obtained so far include the plasma reduction of urea and
the calculated KtN. The ~2-m removal was assessed by percent plasma reduction,
apparent sieving coefficients measured at 5 and 240 min, and total amount in the
collection of total UF volume.
In the first part of the study, PRD-HF was performed at TMP values of 100,
200 and 300 mm Hg. The result is shown in table 1. With the goal of having
maximum removal of urea and ~2-m, we see that TMP should be maximized. The
obtained substitution volumes are high, with a maximum of 88 .7 liters during the
4-h treatment. This corresponds to a filtration rate of more than 370 mllmin. The
resulting urea clearance is sufficient to give KtN values between 1.14 and 1.53.
A Decade of Experience with On-Line HF/HDF
Downloaded by:
Vanderbilt University Library - 10/27/2017 1:59:38 PM
Clinical Results
Table 1. Results of a clinical study comparing PRD-HF at TMP 100, 200 and
300 mm Hg and POD-HF at TMP 400 (PRD-HF at TMP 300 seems to be the most efficient
Urea change
39.0 ~ 12.5
39.0 ~ ILl
38.7 ~ 8.4
29.0 ~ 6.9
28 . 0~
The reduction in the Prm level is similar for the two highest TMPs, while P2-m
recovered in ultrafiltrate is clearly highest at TMP 300. In addition to the P2-m
recovered in ultrafiltrate, some high-flux membranes are known to adsorb significant amounts [20].
The PRD treatment at TMP 300 can then be compared with POD-HF shown
in the last row of table 1. Even though this is an unusually good POD result, it is
far from the PRD treatment. Both urea removal measured by KtN and P2-m
recovered are higher with PRD.
Enough data are now available to state that on-line production is a safe, economic and practical way to supply the large quantities of substitution fluid needed
to achieve the full benefits ofHF and HDF. Furthermore, PRD-HF (and perhaps
HDF) is an interesting alternative to POD-HF and HDF since improved efficiency can be achieved at least for moderate blood flows.
Henderson LW, Besarab A, Michaels A, Bluemle LW Jr: Blood purification by ultrafiltration and
fluid replacement (diafiltration). Trans Am Soc ArtifIntern Organs 1967; 13:216-226.
Baldamus CA, Ernst W, Frei U, Koch KM: Sympathetic and hemodynamic response to volume
removal during different forms of renal replacement therapy. Nephron 1982;31 :324-332.
Downloaded by:
Vanderbilt University Library - 10/27/2017 1:59:38 PM
Quellhorst EA, Schuenemann B, Mietzsch G: Long-term hemofiltration in 'poor risk' patients.
Trans Am Soc ArtifIntern Organs 1987;33:758-764.
Schaefer K, Asmus HG, Quellhorst E, Pauls A, von Herrath D, Jahnke J: Optimum dialysis treatment for patients over 60 years with primary renal disease. Survival data and clinical results from
242 patients treated either by haemodialysis or haemofiltration. Proc Eur Dial Transplant Assoc
Floege J, Granolleras C, Deschodt G, Heck M, Baudin G, Branger B, Toumier 0, Reinhardt B,
Eisenbach GM, Smeby LC, Koch KM, Shaldon S: High-flux synthetic versus cellulosic membranes
for ~2-microg1obulin removal during hemodialysis, hemodiafiltration and hemofiltration. Nephrol
Dial Transplant 1989;4:653-657.
Gejyo F, Odani S, Yamada T, Honma N, Saito H, Suzuki Y, Nakagawa Y, Kobayashi H, Maruyama Y, Hirasawa Y, Suzuki M, Arakawa M: ~2-Microg1obulin: A new form of amyloid protein associated with chronic hemodialysis. Kidney Int 1986;30:385-390.
Jenkins RD, Funk JE, Chen B, Golper TA: Operational instability in extracorporeal filtration of
blood. Blood Purif 1992; 10:292-308.
Sternby J, Jonsson S, Ledebo I: Hemodiafiltration: Technical aspects; in Shaldon S, Koch KM (eds):
Polyamide - The Evolution of a Synthetic Membrane for Renal Therapy. Contrib Nephrol. Basel,
Karger, 1992, vol 96, pp 86-98.
Usuda M, Shinzato T, Sezaki R, Kawanishi A, Maeda K, Kawaguchi S, Shibata M, Toyoda T,
Asakura Y, Ohbayashi S: New simultaneous HF and HD with no infusion fluid. Trans Am Soc Artif
Intern Organs 1982;28:24-27.
Haas T, Uzan M, Pertuiset N, Sacci S, Hillion D: On-line predilutional hemofiltration: Technical
and clinical data (abstract). Blood Purif 1991; 9:25-26.
Nystrand R, Johansson NO: Rejection capacity of microorganisms and endotoxins in an ultrafilter
(abstract). Blood Purif 1984;2:222.
Baz M, Durand C, Ragon A, Jaber K, Andrieu D, Merzouk T, Purgus R, Olmer M, Reynier JP,
Berland Y: Using ultrapure water in hemodialysis delays carpal tunnel syndrome. Int J Artif Organs
Delin K, Attman PO, Dahlberg M, Aurell M: A clinical test of a new device for on-line preparation
of dialysis fluid from bicarbonate powder: The Gambro BiCart. Dial Transplant 1988;17:468482.
Dinarello CA, Lonnemann G, Maxwell R, Shaldon S: Ultrafiltration to reject human interleukinI-inducing substances derived from bacterial cultures. J Clin MicrobioI1987;25: 1233-1238.
Dinarello CA, Maxwell R, Shaldon S: Rejection of human interleukin-I-inducing substances by
ultrafilters (abstract). Blood Purif 1984;2: 197.
Lonnemann G, Mahiout A, Schindler R, Colton CK: Pyrogen retention by the polyamide membranes; in Shaldon S, Koch KM (eds): Polyamide - The Evolution of a Synthetic Membrane for
Renal Therapy. Contrib Nephrol. Basel, Karger, 1992, vo196, pp 47-63.
Schaefer K, von Herrath D, Hartenstein-Koch K, Kutschera D: Clinical experience with a new
in-line hemofiltration system: An analysis of 1,300 treatments. Int J ArtifOrgans 1989; 12(suppl 4):
Canaud B, Imbert E, Kaaki M, Assounga A, Nguyen QY, Stec F, Garred U, Bostrom M, Mion C:
Clinical and microbiological evaluation of a postdilutional hemofiltration system with in-line production of substitution fluid. Blood Purif 1990;8: 160-170.
Rindi P, Pilone N, Ricco Y, Cioni L: Clinical experience with a new hemodiafiltration (HDF)
system. Trans Am Soc ArtifIntern Organs 1988;34:765-768.
Floege J, Smeby LC: Dialysis-related amyloidosis and high-flux membranes; in Shaldon S, Koch
KM (eds): Polyamide - The Evolution of a Synthetic Membrane for Renal Therapy. Contrib Nephrol. Basel, Karger, 1992, vol 96, pp 124-137.
Dr. Jan Sternby, Gambro AB, Box 10101, S-220 10 Lund (Sweden)
A Decade of Experience with On-Line HF/HDF
Downloaded by:
Vanderbilt University Library - 10/27/2017 1:59:38 PM
Без категории
Размер файла
672 Кб
Пожаловаться на содержимое документа