Post-dilution haemodiafiltration and low-flux haemodialysis have dissimilar effects on platelets: a side study of CONTRAST
Mareille Gritters-van den Oever
Muriel P. C. Grooteman
Piet C. M. Bartels
Peter J. Blankestijn
Michiel L. Bots
Marinus A. van den Dorpel
Piet M. ter Wee
Menso J. Nube
Department of Clinical Chemistry, Haematology and Immunology, Medical Centre Alkmaar
Department of Nephrology, VU University Medical Centre
Department of Nephrology, Medical Centre Alkmaar
Executive Committee of CONTRAST
Department of Internal Medicine, Maasstad Hospital
Julius Centre for Health Sciences and Primary Care, University Medical Centre Utrecht
Department of Nephrology, University Medical Centre Utrecht
Background. Cardiovascular disease (CVD) is the leading cause of death in patients with end-stage renal disease (ESRD). Platelet (PLT) dysfunction, which is a wellknown phenomenon in advanced chronic renal failure, corresponds positively with CVD in these patients. The accumulation of retained uraemic toxins might play an important role in this respect. During haemodialysis (HD), both an increase in the expression of the platelet (PLT) cell surface molecule P-selectin (CD62p) and the release of intra-granular substances, such as platelet factor 4 (PF4) and -thromboglobulin (BTG), have been described. As the removal of uraemic toxins is superior during haemodiafiltration (HDF), this form of treatment may have quite another impact on PLTs than HD. Methods. Nineteen chronic HD patients who were treated with low-flux HD for at least 2 months were included in the Dutch CONvective TRAnsport STudy (CONTRAST). After randomization, 10 patients continued low-flux HD and 9 patients switched to post-dilution HDF. The present study describes various parameters of PLT activation and degranulation at baseline (during HD) and after 3 months (during HDF) in the latter group of patients. At both time points, multiple blood samples were drawn. During the first 30 min of treatment, differences over the extracorporeal circuit (ECC) were calculated by taking samples from both afferent (arterial) and efferent (venous) lines. Correlations between various parameters were calculated in the total group of patients after 3 months. Results. Immediately after the start of HD, PLT counts dropped over the ECC. During HDF, PLT counts decreased even more and reached a nadir at t30. CD62p expression increased early during HD and returned to baseline thereafter. During HDF, these changes were more pronounced and more protracted. With respect to degranulation, rather dissimilar results were obtained. During HD, both PF4 and BTG increased over time, whereas during HDF, PF4 inC The Author 2009. Published by Oxford University Press [on behalf of ERA-EDTA]. All rights reserved. For Permissions, please e-mail:
creased but BTG did not change. Haemoconcentration and
transmembrane pressure (TMP) within the dialyser were,
respectively, 10 and 3 higher during HDF than during
HD. There was a striking correlation between the changes
in haemoconcentration and the changes in both PLT counts
and CD62p over the ECC.
Summary and Conclusions. PLT activation, as measured
by the expression of CD62p, was more pronounced and
more protracted during HDF than during HD. During HDF,
PLTs were trapped abundantly within the ECC, not only
after first passage, but also thereafter. The degranulation
product BTG increased during HD, but did not change
during HDF. These observations may well be explained by
the greater haemoconcentration and/or higher TMP during
HDF on the one hand, and superior convective transport at
the other. Whether the potential harmful effects of enhanced
PLT activation are counterbalanced by the beneficial effects
of an increased convective transport of degranulation
products remains to be established.
During haemodialysis (HD), various side effects occur,
including the activation of protein systems in the blood and the
stimulation of circulating elements. The sum of these
undesirable side effects has been termed bio-incompatibility
. With respect to platelets (PLT), both an increase
in the expression of cell surface molecules and the
release of intra-granular substances, such as platelet
factor 4 (PF4), platelet-derived growth factor (PDGF) and
-thromboglobulin (BTG), have been well documented .
During HD, activated PLTs adhere to circulating elements,
including other PLTs, neutrophils, lymphocytes ,
monocytes and erythrocytes . As a result of these cellcell
interactions, a mixture of micro-aggregates is formed, with
specific pathophysiological effects. Flow cytometric
analysis showed production of reactive oxygen species after the
intra-dialytic formation of plateletneutrophil aggregates
. In the clinical situation, PLT reactivity corresponded
positively with cardiovascular morbidity and mortality, both
in non-renal patients  and in patients with end-stage
renal disease (ESRD) . Moreover, in HD patients, PLT
activation correlated with bleeding and thrombotic
diathesis  and vascular access failure . The cause of PLT
dysfunction in patients with ESRD is multi-factorial. Both
uraemic and HD-related factors might play a role. Not only
the type of dialyser  but also other components of the
extracorporeal circuit (ECC)  and the type of
anticoagulant  influence PLTs. As PLT abnormalities occur
early in the course of renal failure , the uraemic
milieu and particularly the retention of uraemic toxins
probably also have a profound influence on these elements .
Therefore, dialysis techniques with a high convective
transport, such as haemodiafiltration (HDF), may reduce the
degree of PLT activation and improve PLT reactivity.
However, as a high convection volume can only be reached at
the cost of substantial haemoconcentration, at least
during post-dilution HDF, and an increased transmembrane
pressure (TMP) within the dialyser, it is conceivable that
undesirable side effects occur. The present analysis is a
single-centre side study of CONTRAST, the Dutch
multicentre randomized controlled trial comparing the effects
of post-dilution HDF with low-flux HD on all-cause
mortality and cardiovascular morbidity and mortality . At
baseline and 3 months after randomization, PLT numbers,
the expression of the cell surface marker CD62p and the
degranulation products PF4 and BTG were measured. In
order to establish the role of the ECC in PLT activation and
degranulation, blood was sampled from both afferent
(arterial) and efferent (venous) lines at various time points during
Subjects and methods
In our centre, 19 stable ESRD patients [11 males and 8 females,
median age 63 years (2882)] undergoing low-flux HD treatment for at least
2 months [median 25 months (783)] participated in CONTRAST. Of
these patients, 10 were randomized to treatment with low-flux HD and
9 to post-dilution HDF. In the latter group, PLT parameters were
measured prospectively before (during HD) and after the switch to HDF at
3 months, thus excluding patient and non-HD(F)-treatment-related
factors as much as possible. Correlations between various parameters were
calculated cross-sectionally by combining the data of the total group of
patients (n = 19) 3 months after randomization. The aetiology of
renal insufficiency was hypertensive nephrosclerosis in eight patients,
diabetic nephropathy in five and adult dominant polycystic kidney disease
in three patients. The other three patients suffered from IgA
nephropathy, membranous nephropathy and tubulo-interstitial nephritis.
Individuals who used platelet inhibitors or other medication interfering with PLT
function were excluded from this study. EPO dose [median 10 000 (2000
30 000 IU)] per week remained unchanged over the study period. Written
informed consent was obtained in all cases. The protocol was approved
by the local Medical Ethical Committee and the Executive Committee of
All parameters were measured over time both prior to randomization (M0)
and after 3 months (M3). Samples were taken from the fistula (t0) and
from the efferent line, i.e. after the dialyser and before the addition of
substitution fluid in HDF, after 1 (t1), 5 (t5), 30 (t30), 60 (t60) and 150
(t150) min. Furthermore, samples were drawn from the afferent lines,
i.e. before the roller pump, after 5 min (t5aff) and 30 min (t30aff). The
calculated differences over the ECC (delta, = efferent value afferent
value) at t1, t5 and t30 were designated as t1 (first passage), t5 and
Platelet numbers were corrected for changes in plasma volume
based on haematocrit (Ht) measurements: [corrected valuetx = valuetx
(Htt0/Httx)]. All other results were corrected for changes in plasma
volume, based on a different, more complex, formula: corrected valuetx =
valuetx [Htt0/(1 Htt0)][(1 Httx)/Httx].
Haemodialysis procedure and materials
Only first-use low-flux polysulfone (PS) dialysers [F8 HPS,
Fresenius Medical Care, Bad Homburg, Germany; ultrafiltration (UF) factor
18 ml/h mmHg, surface area 1.8 m2, steam sterilized] were utilized on
Fresenius 4008 and 5008 series dialysis machines. Bicarbonate dialysate
was used with a dialysate flow of 500 ml/min, whereas dialysate
temperature was kept at 36C. The production of ultrapure water was similar to
HDF (see below). All dialysers were pre-rinsed with 1000 ml 0.9% NaCl.
Individual doses of dalteparin were based on bodyweight (50 IU/kg) and
given as a bolus injection at the beginning of the dialysis session (mean
4750 1419 IU). According to the CONTRAST protocol, treatment times
were fixed at baseline and spKt/V urea was 1.2 per treatment.
Haemodiafiltration procedure and materials
Only first-use high-flux PS dialysers [FX80 Helixone R , Fresenius
Medical Care, Bad Homburg, Germany; ultrafiltration (UF) factor (Kuf)
59 ml/h mmHg, surface area 1.8 m2, steam sterilized] were used. A
sterile and non-pyrogen substitution solution was prepared online from
a dialysate concentrate and ultrapure dialysate using the ONLINEplustm
system (Fresenius Medical Care), which contains two ultrafilters
(DIASAFE R plus) and is integrated in the Fresenius 4008 and 5008 series
dialysis machines. Ultrapure quality was defined as bacterial counts
<0.1 CFU/ml and endotoxin levels <0.025 EU/ml. The microbiological
water quality of the dialysis solutions was regularly monitored according
to the guidelines of the Dutch Federation of Nephrology. Bicarbonate
was provided from powder cartridges (biBAG, Fresenius Medical Care).
The dialysate flow was kept at 500 ml/min, with a dialysate temperature
of 36C. According to the CONTRAST protocol, the blood flow was
250400 ml/min, treatment times were fixed and spKt/V urea was 1.2
per treatment. HDF was provided in the post-dilution modus with a target
convection volume of 6 l/h. All dialysers were pre-rinsed with 1000 ml
0.9% NaCl. The dalteparin protocol was similar to HD, although in the
first months after randomization two patients were switched to higher
doses (mean 5250 1845 IU, n = 9) following an episode of coagulation
in the dialyser.
Transmembrane pressure (TMP) and haemoconcentration
As the TMP is not displayed on the dialysis machines, an estimation was
made based on dialyser membrane characteristics (Kuf) and ultrafiltration
(UF) rate (Quf) according to the formula: TMP = Quf/Kuf. In addition,
the UF pressures, which are shown on the display of the dialysis machines,
were noted throughout the HD and HDF sessions and served as a surrogate
marker of TMP. Both filtration fraction (FF) and haemoconcentration
(% increase in hematocrit) in the ECC were calculated based on Ht levels
in the afferent and efferent lines at t5 and t30.
Platelet counts. Blood samples were collected in K2EDTA (ethylene
diamine tetraacetic acid) tubes (Becton Dickinson, Plymouth, UK) and
PLT numbers were determined using a Sysmex XE2100 Haematology
analyser (Sysmex Corporation, Kobe, Japan).
Platelet surface markers. The platelet surface markers CD62p
(P-selectin; clone CLB Thromb 6, Beckman Coulter, Mijdrecht,
The Netherlands) and CD41 (clone P2, Beckman Coulter, Mijdrecht,
The Netherlands) were detected by direct labelling and flow cytometry.
Blood was drawn into K2EDTA tubes and within 2 h after collection
incubated with a glycoprotein-specific fluorochrome-labelled monoclonal
antibody. A flow cytometer (Epics XL, Beckman Coulter, Mijdrecht, The
Netherlands) was used to determine the percentage of platelets with CD62p
surface expression. CD41 served as a platelet-specific label.
Platelet factor 4 and beta-thromboglobulin. Blood samples were drawn
into CTAD tubes (Vacutainer R CTAD, Becton Dickinson, Plymouth, UK),
cooled on ice and centrifuged for 20 min at 2500 g. Plasma samples were
stored at 70C until measurement. PF4 and BTG were determined using
commercially available sandwich ELISA kits (Asserachrom PF4 R and
Asserachrom b-TG kit R , Diagnostica Stago, Asnie`res, France).
All analyses were performed with the SPSS 15.0 software system. A
general linear model (GLM) for repeated measures was applied to study
various aspects of platelet activation during a single session both at
baseline (HD) and 3 months after randomization (HDF). The GLM was
applied as it allows the simultaneous assessment of the effects of both
time and treatment modality on outcome measures. Furthermore, GLM
was used to analyse the effect of treatment modality and sampling point
[efferent versus afferent, delta () over the ECC] on the same outcome
parameters. In the case of significance, post hoc paired t-tests were
applied. Finally, correlations between haemoconcentration, PLT numbers
and CD62p were calculated and expressed as Pearsons coefficients. Data
were expressed as mean standard error. Differences were considered
significant at P < 0.05.
Baseline (M0) versus 3 months (M3). PLT counts at M3
t0 were not different from M0 t0 (data not shown).
Intra-dialytic changes over time (t0 t150). As shown
in Figure 1A, during HD PLT counts changed only slightly
over time: from 196 12 109/l at t0 to 180 11 109/l
at t30 and 188 13 109/l at 150 (P = 0.010). In the HDF
group, after an initial decline from 195 13 109/l at t0 to
162 11 109/l at t30 (efferent line), the number of these
cells increased substantially to 233 17 109/l at t150
(P < 0.001). The difference between the two modalities
was highly significant (P < 0.001). In the HDF group, both
at t5 and at t30 a marked increment was observed in the
afferent line (Figure 1B).
Changes over the ECC (t1, t5 and t30). As shown
in Figure 1C, during HD PLT counts declined only at t1,
i.e. after first passage (t1 31 11 109/l, P = 0.021).
During HDF, however, PLTs declined significantly over the
circuit not only at t1 (t1: 45 16 109/l), but also at
t5 (t5: 53 13 109/l) and t30 (t30: 77 12
109/l). The difference between HD and HDF was highly
significant, both at t5 and at t30 (P = 0.009).
Baseline (M0) versus 3 months (M3). CD62p expression
at M3 t0 was not different from M0 t0 (data not shown).
Intra-dialytic changes over time (t0 t150). During HD,
CD62p expression increased markedly: from 29 6% at
t0 to 48 8% at t30 and declined thereafter to 30 4% at
t150 (P < 0.001, Figure 2A). During HDF, however, CD62p
expression increased and remained elevated until the end
of treatment (t150: 55 8%, P < 0.001). The difference
between HD and HDF was highly significant (P = 0.002).
Changes over the ECC (t1, t5 and t30). During both
HD and HDF, the expression of CD62p increased
significantly over the ECC at t1, t5 and t30 (Figure 2B). The
difference between HD and HDF was significant (P =
0.039) at t30.
Platelet factor 4
Baseline (M0) versus 3 months (M3). PF4 levels at M3 t0
were not different from M0 t0 (data not shown).
Intra-dialytic changes over time (t0 t150). During both
HD and HDF, an increase in PF4 levels was observed over
time (P < 0.001, Figure 3A), reaching maximum values
at t5 (HD: 107 10 IU/ml, HDF: 84 3 IU/ml). At
t150, in both HD and HDF, PF4 levels were still elevated if
compared to t0. No differences were observed between the
two treatment modalities.
Changes over the ECC (t1, t5 and t30). During HD,
PF4 increased across the ECC (t1: P = 0.005, t30: P =
0.046, Figure 3B), whereas PF4 concentrations decreased
over the ECC during HDF (t5: P = 0.021). The difference
between HD and HDF was highly significant (P = 0.001).
Baseline (M0) versus 3 months (M3). BTG levels at M3
t0 were not different from M0 t0 (data not shown).
Intra-dialytic changes over time (t0 t150). During HD,
a significant increase was observed from 214 46 IU/ml
at t0 to 361 51 IU/ml at t30, with a decline thereafter to
281 33 IU/ml at t150 (P = 0.013, Figure 4A). During
HDF, changes were not observed. The difference between
HD and HDF was significant (P = 0.03).
Changes over the ECC (t1, t5 and t30). During
HD, BTG increased moderately over the ECC (t30: P =
0.06, Figure 4B). During HDF, marked changes were not
observed. No differences were observed between the two
Transmembrane pressure and haemoconcentration
Based on the formula (TMP = Quf/Kuf), the estimated
TMPs in HDF were 100 mmHg and <50 mmHg in HD.
UF pressures as measured and displayed by the machine
ranged from 120 to 260 mmHg during HDF, whereas in
HD these pressures were generally <100 mmHg and mostly
<50 mmHg. The correlation between the estimated TMP
and the UF pressures was highly significant (r2 = 0.8,
P = 0.01). Hence, it appears that TMP during HDF is 3
higher than during HD.
The filtration fraction, as calculated from Ht levels in the
afferent and efferent lines at t5 and t30 based on the formula
[1 (Htaff/Hteff)], was 0.03 0.01 in HD and 0.29 0.01
Fig. 1. (A) PLT numbers (109/l, mean SE) in the efferent line during HD and HDF. Changes over time were observed during both treatments:
(HD) P = 0.010, + (HDF) P < 0.001. PLT counts evolved differently over time during treatment with both techniques (P < 0.001). (B) PLT
numbers (109/l, mean SE) in the both afferent and efferent lines during HDF: P = 0.038, P = 0.002. NB. P-values for changes over the ECC
are depicted in C. (C) Changes in PLT numbers (109/l, mean SE) over the ECC (delta: efferent value afferent value) during HD and HDF: P <
0.05. Difference between HD and HDF: P = 0.009.
Fig. 3. (A) PF4 concentrations (IU/ml, mean SE) in the efferent line during HD and HDF. Changes over time were observed during both treatments:
(HD)/+ (HDF): P < 0.001. (B) Changes in PF4 concentrations (IU/ml, mean SE) over the ECC (delta: efferent value afferent value) during HD
and HDF: P < 0.05. Difference between HD and HDF: P = 0.001.
Fig. 4. (A) BTG (IU/ml, mean SE) concentrations in the efferent line during HD and HDF. Changes over time were observed during HD: P =
0.013. Difference between HD and HDF: P = 0.030. (B) Changes in BTG (IU/ml, mean SE) concentrations over the ECC (delta: efferent value
afferent value) during HD and HDF.
in HDF, and the resulting haemoconcentrations [(Hteff
Htaff/Htaff) 100%] 4 1% and 41 1%, respectively.
Correlations between the haemoconcentration (%) within
the ECC and platelet numbers and activation markers
The combined data from both HD and HDF group at
M3 (n = 19) at t30 showed a robust correlation
between the haemoconcentrationand likewise the
filtration fractionand the changes in both CD62p expression
(t30, r2 0.46, P = 0.002, Figure 5) and PLT numbers over
the ECC (t30, r2 0.68, P < 0.001).
PLT activation and degranulation are well-known side
effects of HD, which may aggravate the PLT dysfunction that
is already observed in the preceding stages of CRF [8,15].
Hence, in ESRD, both bio-incompatible effects of
repetitive HD and accumulation of uraemic toxins may have a
damaging impact on PLTs. Activated PLTs release
inflammatory and mitogenic substances that influence
endothelial cell function, chemotaxis and adhesion, and promote
the transmigration of monocytes to sites of damaged
endothelium, as in early atherosclerotic plaques . In cell
cultures, it was demonstrated that the degranulation
product PF4 binds directly to oxidized LDL, thereby increasing
its binding to vascular cells and macrophages , thus
enhancing the formation of foam cells and contributing to
the process of atherosclerosis. Hence, it has been suggested
that HD-induced PLT activation contributes to the vascular
damage and accelerated atherosclerosis, which is observed
in patients with ESRD .
In the present study, the activation and degranulation
of PLTs were estimated prospectively in patients who
were first treated with low-flux HD and thereafter with
Fig. 5. Correlations between the haemoconcentration (% increase in
haematocrit over the ECC) and changes in platelet numbers (10E9/l,
r2 0.68) and CD62p expression (% PLT, r2 0.46) over the ECC at t30
(delta: efferent value afferent value). Combined data from both HD and
HDF groups at M3 (n = 19).
post-dilution HDF. As the material of the dialysers was
similar in both treatment arms (polysulfone), at first sight
the main difference between the two modalities is the
amount of convective transport. Theoretically, PLT
dysfunction could improve by HDF, as uraemic retention
products play an important role in this respect . However, as
demonstrated in this study, haemoconcentration increased
by 40%, whereas TMP was 3 higher than in HD. In
fact, both these effects may have an adverse influence on
PLTs that traverse the dialyser in HDF patients.
During HD, PLT counts dropped over the ECC, but only
directly after the start of treatment. During HDF, however,
PLT counts reached their nadir at t30, indicating that PLTs
are trapped within the ECC not only after first passage, but
also thereafter. In the HDF group, both at t5 and at t30 PLT
numbers were higher in the afferent line, indicating the
release of PLTs from the bone marrow or storage pools.
Furthermore, a substantial rebound was noticed at t150.
Together, these data suggest that fresh PLTs are
continuously released during HDF, which is quite different from
the situation during HD. As samples taken before treatment
at baseline and after 3 months were similar, these data
indicate that PLT numbers return to baseline within days.
During low-flux HD, CD62p expression increased over the
ECC immediately after the start and returned to baseline at
t150. During HDF, however, the alterations were more
pronounced and more protracted. Moreover, whereas the
changes over the ECC during HD at t1, t5 and t30 remained
relatively stable, these values gradually increased during
HDF. Samples taken before treatment at baseline and after
3 months showed no differences, neither in HD nor in HDF.
Hence, the combined data on PLT numbers and CD62p
expression suggest that activated PLTs disappear from the
circulation before the next treatment and are replaced by an
influx of fresh PLTs from storage pools. Although we did
not assess TMP directly by a four-point pressure
measurement, a highly significant correlation was found between
the calculations of the formula TMP = Quf/Kuf and the
UF-pressures, as assessed by a two-point (4008) or
threepoint (5008) pressure measurement, which are displayed
on the machines. The estimated TMPs were 3 higher
in HDF than in HD, whereas in HDF, with a mean FF of
0.29, the haemoconcentration over the dialyser was 40%.
Therefore, in our opinion, the above-mentioned differences
between HD and HDF may very well be attributable to
a dissimilar haemoconcentration and/or TMP, leading to
different flow and shear conditions within the dialyser
during treatment . The former conclusion is obviously
supported by the striking correlation between the
haemoconcentration observed and the deltas of both PLT counts
and CD62 expression in the total group of patients at M3.
With respect to PLT degranulation, PF4 levels reached
peak values shortly after the start of treatment and decreased
gradually thereafter, during both HD and HDF. Comparable
findings were published before, both in low-flux  and in
high-flux HD . As PF4 increased particularly in the
afferent line, and the changes over the ECC were only modest
(HD) or even absent (HDF), these findings imply that the
rise in PF4 occurred mainly outside the ECC. Moreover,
these data suggest loss of PF4 from the blood compartment
through the dialyser to the dialysate compartment during
HDF. Previously, we described a sharp increment in PF4
after the isolated administration of low-molecular-weight
heparin (LMWH) 10 min before the actual start of HD
. From these data, it was concluded that this effect is
mainly caused by LMWH-induced detachment of PF4 from
the endothelium and only to a limited extent by release from
PLTs. Clearly, this view is confirmed and strengthened
by the present data. Hence, other PLT granule products,
which are not released by LMWH from the endothelium,
may be more appropriate for the estimation of HD-induced
PLT degranulation. Indeed, recently we demonstrated that
BTG, which is also stored in the -granules of PLTs, is
almost exclusively released within the ECC . Likewise,
in the present study, BTG levels increased markedly over
the ECC during HD, which is in line with other studies
in low-flux HD [2,21]. During HDF, however, BTG levels
did not change, neither in the ECC, nor over time. As the
molecular weights (MWs) of PF4 and BTG are 27 kD and
36 kD, respectively, it seems plausible that these substances
are removed by convective transport, which is obviously not
the case in low-flux HD. Currently, it is unclear whether,
and to what extent, BTG is removed by convective transport
during high-flux HD. In two studies comparing regenerated
cellulose (RC) and cellulose triacetate (CTA) with
polysulfon (PS) membranes, BTG levels increased significantly
during HD with RC and CTA, but not with PS [22,23].
However, in these two studies, the UF characteristics of the
dialysers were not specified. Although PLTs in HD patients
appear hypo-responsive to stimulation , exhaustion due
to the repetitive stimulation of the treatment  is unlikely
in the present analysis. In both modalities, a substantial
PF4 increase was observed in the venous line directly after
first passage, which clearly could not originate from the
endothelium but only from PLTs traversing the ECC.
Obviously, our study has some limitations. First, this
analysis was a side study of CONTRAST, comparing
PLT activation during online post-dilution HDF to regular
low-flux HD. Since a high-flux filter was applied in
HDF, the difference between the two modalities
comprised not only convective transport related matters (i.e.
haemoconcentration and higher TMPs), but also membrane
flux characteristics. Hence, it is not readily apparent from
our data whether the dissimilar patterns of PLT activation
and degranulation result from differences in the
permeability characteristics of the membranes used and/or the
dissimilar flow and shear conditions within the dialysers.
As the CONTRAST study protocol precluded the
inclusion of other treatment modalities, such as pre-dilution HDF
or high-flux HD, we can only speculate on the effects of
membrane permeability. According to the formula TMP =
QUF/KUF, the larger pore size (higher KUF) of high-flux
membranes would result in lower TMPs, whereas
haemoconcentration would be greater due to internal filtration.
In this respect, it should be mentioned that PLT activation
during high-flux HD [4,10] was not different from
lowflux HD, suggesting that a pore size alone has no major
effect on this process. Another shortcoming in this study
concerns the degree of anticoagulation. As anti-Xa
activity is not monitored routinely in our centre, the dosage of
dalteparin was based on body weight and adjusted in the
case of clotting of the ECC. Considering the molecular
weight of dalteparin, which is 5000 D, it is likely that
part of it is removed during HDF . Moreover, recently
it appeared that procoagulatory activity was higher during
pre-dilution HDF than during high-flux HD, despite
comparable anti-Xa activity . Therefore, it cannot be excluded
that in the present study, during HDF some unnoticed
clotting occurred within the dialyser, which altered local flow
conditions and contributed to PLT activation.
Finally, our data only refer to post-dilution HDF, which
is quite different from pre-dilution HDF, in particular with
respect to the degree of haemoconcentration and
possibly coagulation. On the other hand, as all patients were
treated according to standard treatment protocols of the
Dutch Federation of Nephrology and the HDF protocol
of CONTRAST is consistent with current guidelines, our
study is representative for todays clinical practice.
To conclude, PLT activation, as measured by the
expression of CD62p, is more pronounced and more protracted
during post-dilution HDF than during low-flux HD. During
HDF, PLTs are trapped in large quantities within the ECC,
not only after first passage, but also thereafter. Moreover,
as PLT counts in the afferent line increased over time and
a marked rebound was observed during the course of
treatment, it is likely that fresh PLTs are continuously released
from the resident pool. As CD62p expression did not
return to baseline, it seems plausible that these fresh PLTs
are activated during the entire course of treatment.
During HD, these phenomena are either absent or much less
prominent. Whereas mainly quantitative differences were
observed with respect to PLT counts and CD62p
expression, the levels of degranulation products were markedly
lower during post-dilution HDF as compared to low-flux
HD. As the MWs of both PF4 and BTG are clearly
below the cut-off value of the high-flux dialysers used, it is
conceivable that these substances are cleared from the
circulation by convective transport. The different kinetics of
BTG and PF4 may be explained by the fact that BTG is
exclusively released within the ECC and PF4 originates
mainly from the endothelium. Hence, in the case of PF4,
only a fraction of the total body burden crosses the ECC
per unit of time, whereas in the case of BTG all
production is directly subjected to convective transport. Due to the
fact that neither PF4, nor BTG, traverses low-flux
membranes, the amount of both products increased during HD.
The observations during HDF may well be explained by
the greater haemoconcentration and/or higher TMP on the
one hand, and superior convective transport at the other.
Whether the potential harmful effects of enhanced PLT
activation are counterbalanced by the possible beneficial
effects of increased convective transport and hence removal
of degranulation products is currently unknown.
Acknowledgements. We wish to thank the patients of the Medical Centre
Alkmaar Dialysis Department for their participation in this study and the
nursing staff for their indispensable support. We are grateful to the Dutch
Kidney Foundation, Fresenius Medical Care and Gambro for their
financial support to CONTRAST. The present study was financially supported
by unrestricted grants from Baxter and Fresenius Medical Care.
Conflict of interest statement. The sponsors were neither involved in the
analysis of the results, nor in the writing of the manuscript.
Received for publication: 7.1.09; Accepted in revised form: 3.6.09
Qunying Guo2, Juan Jes us Carrero1, Xueqing Yu2, Peter Barany1, Abdul Rashid Qureshi1,
Monica Eriksson1, Bj orn Anderstam1, Michal Chmielewski1, Olof Heimb urger1, Peter Stenvinkel1,
Bengt Lindholm1 and Jonas Axelsson1
Background. Vascular endothelial growth factor (VEGF)
was recently shown to predict survival in
prevalent haemodialysis patients. Soluble VEGF receptors
(sVEGFR)-1 and -2 are circulating endogenous modulators
of VEGF activity. We thus studied the relationship between
sVEGFR-1 and -2 and survival in a cohort of prevalent
haemodialysis (HD) patients.
Methods. Components of the VEGF system were measured
(ELISAs) in 185 prevalent HD patients and levels related to
clinical characteristics, biochemical markers and survival.
The patients were followed up prospectively for a median
31 (2037) months.
Results. While ischaemic heart disease was independently
associated with a lower sVEGFR-2 (OR = 2.75, P = 0.02),
sVEGFR-1 was positively associated with IL-6 ( = 0.22,
P = 0.003) and white blood cell count ( = 0.22, P = 0.002).
In survival analysis, the patients with a high sVEGFR-1
level had a higher all-cause mortality (KaplanMeier
ChiSquare = 5.6, P = 0.02) and a higher adjusted mortality
risk (Cox HR = 1.93, P = 0.009) than those with low
Conclusion. In the first clinical study of sVEGFR-1 and -2
in CKD, we found novel associations between the sVEGFRs
and cardiac disease. This may be of clinical importance, as
a high sVEGFR-1 was an independent risk factor for