- Email: [email protected]
Effect of Renal Replacement Therapy on Viscosity in End-Stage Renal Disease Patients
Effect of Renal Replacement Therapy on Viscosity in End-Stage Renal Disease Patients Mariano Feriani, MD, Paul L. Kimmel, MD, Joseph Kurantsin-Mills, PhD, and Juan P. Bosch, MD • Viscosity, an important determinant of microcirculatory hemodynamics, is related to hematocrit (HCT), and may be altered by renal failure or its treatment. To assess these factors, we studied the effect of dialysis on the viscosity of whole blood, plasma, and reconstituted 70% HCT blood of eight continuous ambulatory peritoneal dialysis (CAPO) and nine hemodialysis (HO) patients under steady shear flow conditions at different shear rates, before and after dialysis, compared with nine normal subjects. The density of the red blood cells (RBCs), a marker of cell hydration, was measured in HO patients by a nonaqueous differential floatation technique. Whole blood viscosity was higher in controls than patients, and correlated with HCT before treatment (P < 0.05) at shear rates of 11.5 to 230 S- 1 in HO patients, and 23 to 230 S-1 in all end-stage renal disease (ESRO) patients. In contrast, whole blood viscosity correlated with HCT in CAPO patients only at the lowest shear rates (2.3 and 5.75 s- \ P < 0.05). Plasma viscosity was higher in CAPO patients than both HO patients before treatment and controls (P < 0.05, analysis of variance [ANOVA]), despite lower plasma total protein, albumin, and similar fibrinogen concentration compared with HO patients. When all samples were reconstituted to 70% HCT, CAPO patients had higher whole blood viscosity than control subjects'. The high HCT blood viscosity of the ESRO patients was higher than control subjects' at capillary shear rates, suggesting increased RBC aggregation and decreased RBC deformability in patients with renal disease. Whole blood viscosity increased concomitantly with ultrafiltration during HO treatment, and was similar in control subjects and HO patients after treatment, although patients had lower HCT than the controls (P < 0.001). Plasma viscosity was higher in HO patients after dialysis compared with controls, due to increased plasma protein concentration. The density profile of the erythrocytes increased after HO, and their median density increased linearly as a function of the ultrafiltered volume. We conclude that uremia per se significantly alters blood viSCOSity, independently of HCT. Patients with ESRO treated with CAPO have higher plasma viscosity despite lower plasma protein concentration, and higher whole blood viscosity at increased HCT levels. Alterations in total plasma protein and fibrinogen concentration, and the effect of ultrafiltration on these parameters, may play critical roles in determining hyperviscosity in patients treated with HO. Changes in RBC surface area to volume ratio and increased RBC density, and hence internal viscosity, as well as elevated plasma protein levels, may induce RBC aggregation. These cellular alterations may affect microcirculatory hemodynamics and predispose patients with ESRO to coagulation abnormalities. © 1992 by the National Kidney Foundation, Inc. INDEX WORDS: Viscosity; erythrocyte density; continuous ambulatory peritoneal dialysis; fibrinogen; hematocrit; hemodialysis.
W
HOLE BLOOD viscosity is a function of plasma viscosity and hematocrit (HCT) levels. I Increased blood viscosity has been associated with increased peripheral vascular resistance, hypertension, and cardiac hypertrophy. I Whole blood viscosity is increased in patient populations with renal disease and hypertension. 2-4 The mechanisms underlying the hemodynamic changes in patients with renal disease are not well understood. However, a number of studies have demonstrated a direct relation between HCT and vascular resistance. 3,5-7 Other factors such as increased levels of fibrinogen may be associated with hyperviscosity in patients with chronic renal insufficiency, and patients with end-stage renal disease (ESRD) treated with hemodialysis (HD).4 Treatment of patients with ESRD by HD, intermittent peritoneal dialysis, or continuous ambulatory peritoneal dialysis (CAPD) results in the reduction of plasma volume, improvement of anemia, and clearance of uremic toxins. Although anemia is a problem
in the ESRD population, hemoconcentration may occur in patients, coincident with net ultrafiltration. Whole blood and plasma viscosity in patients with ESRD treated with HD has been shown to increase after treatment, associated with weight loss and increase in HCT level. 8 While viscosity is a function of HCT, and its alterations might From the Department o/Nephrology, St. Bortolo Hospital, Vicenza, Italy; and the Department 0/ Medicine, George Washington University Medical Center, Washington, DC. Supported in part/rom/unds provided by the National Institutes 0/ Health (Grant No. HL-38521 to J K-M) and the Department 0/ Medicine, George Washington University Medical Center. Presented in part at the 23rd Annual meeting a/the American Society a/Nephrology, Washington, DC, December 1990. Address reprint requests to Paul L. Kimmel, MD, Division 0/R enal Diseases and Hypertension, Department a/Medicine, George Washington University, 2150 Pennsylvania Ave, NW, Washington, DC 20037. © 1992 by the National Kidney Foundation, Inc. 0272-6386/92/1902-0004$3.00/0
American Journal of Kidney Diseases, Vol XIX, No 2 (February), 1992: pp 131-139
131
FERIANI ET AL
132
contribute to increased peripheral vascular resistance, its specific role, if any, in affecting mortality in such patients is unknown. The recent successful use of recombinant DNA human erythropoietin (rHuEPO) to treat the anemia of ESRD has raised concern about the development or worsening of hypertension in a population of treated ESRD patients because of the increased HCT and the associated elevated blood viscosity.· Several studies9 - •• demonstrated increases in whole blood viscosity over time in HD patients treated with rHuEPO; however, these findings have been disputed. 12 Vaziri et al9 showed an increase in whole blood viscosity in rHuEPO-treated HD patients when postdialysis treatment values were compared with baseline values. Finally, few studies of viscosity in patients with ESRD treated with CAPD have been performedY Comparisons between viscosity measurements before and after HD treatment with values in patients treated with CAPD have not been made. To assess the rheological factors associated with ESRD and its treatment, and to compare the effects of different renal replacement therapies, we investigated the effect of dialysis on whole blood and plasma viscosity in patients treated with CAPD and HD under steady shear flow conditions, and compared the data with those of normal healthy subjects. METHODS
Patients Nine patients with ESRD treated with HD, aged 36 to 79 years (mean, 60 ± 5), volunteered for the study. Three patients had diabetic nephropathy, three had hypertensive nephropathy, and one each had obstructive nephropathy, chronic glomerulonephritis, and unknown renal disease. AU patients were either dialyzed with high-efficiency or high-flux dialyzers, three times weekly. Weight was measured before and after each treatment. All patients were clinically stable at the time of investigation and had been dialyzed for at least I year. Whole blood samples were drawn from the arterial side of the circuit just after the beginning of dialysis, and just before the end of dialysis_Blood was obtained in all cases by venipuncture, using disodium ethylenediamine-tetraacetate (EDTA-2Na) I mg! mL whole blood as an anticoagulant. Eight patients with ESRD treated with CAPD, aged 23 to 62 years (mean, 36 ± 5), were studied simultaneously. Two patients had diabetic nephropathy, three had hypertensive nephropathy, two had unknown renal disease, and one had chronic renal failure complicating horseshoe kidney. Blood samples were obtained at the time of monthly routine visits. Nine healthy subjects, aged 30 to 60 years (mean, 39 ± 4), served as a control population.
All patients gave their informed consent before participating. The study was approved by the George Washington University Medical Center Committee on Human Research.
Determination of Viscosity Measurements of viscosity were made on I-ml aliquots of normal human autologous plasma (as controls) and l-mL aliquots of whole blood of the patients. The cell preparations were maintained at a constant temperature of 37°C. All viscosity measurements were performed within 2 hours of blood collection, in triplicate for each sample. Intrasample coefficient of variation was approximately 5%. To assess the effect of red blood cell (RBC) concentration on viscosity, the whole blood was also adjusted to approximately 70% HCT for the control and patient samples using autologous plasma, and used to measure the viscosity. This required centrifuged separation of the blood for 15 minutes at 2,000 X g, 25°C, and resuspension of the cells in the autologous plasma only. Aliquots (I ml) of the resuspended high HCT blood of control samples or 1ml aliquots of samples of ESRD patients were used for the viscosity measurements. Viscosity was measured in a WellsBrookfield model LVT cone-plate microviscometer (Brookfield Engineering Labs, Stoughton, MA) at a constant temperature of 37°C. Shear rate, calculated from rotation speed and the geometry of the cone and plate, varied from 230 to 2.3 S- I. Shear stress was calculated from the torque (T in dyne centimeters) acting on the cone of the viscometer. 14.15 Apparent viscosity (1/, hereafter referred to as viscosity) is defined as the ratio shear stress to shear rate in dyne· s/cm 2 or poise: viscosity = shear stress (dynes/cm 2)/shear rate (S-I). The result multiplied by 100 yields the unit centipoise (cp).
Determination of RBC Density Profiles The density distribution profile of the RBCs was determined before and after treatment in HD patients, using the phthalate ester density technique. The details of the technique have been described elsewhere. 16 Briefly, the density distribution profiles of the various cell suspensions were analyzed in phthalate esters prepared from 20 mixtures of di-n-butyl phthalate (specific gravity, 1.042) and dimethyl phthalate (specific gravity, \.189) (Fisher Scientific, Pittsburgh, PAl with final specific gravity of the mixtures between 1.062 and 1.138 in increments of 0.004. Duplicate 60-IlL samples of the blood preparations were pipetted into microcapillary tubes (75-mm long, 1.15mm internal diameter) containing 10 ILL of each of the phthalate ester mixtures. The end of the microcapillary tube in contact with the phthalate was sealed by flaming. The microcapillary tubes were then centrifuged at 13,460 X g, for 15 minutes, during which time the temperature equilibrated at 36.5°C (IEC MC Microhematocrit Centrifuge, Needham Heights, MA). The distribution of RBCs was analyzed by their differential floatation on the esters. The percentage of RBCs below the ester after the equilibrium centrifugation was related to the ester density using a computer program (PEDDS; Division of Hematology-Oncology, George Washington University Medical Center). The density profile curves generated from the data were analyzed for numerical indices that quantitatively described the properties of the cell population. These indices are the median density (D50), the density of the phthalate ester at which 50% of the cells are located after equi-
VISCOSITY IN HEMODIALYSIS AND CAPD PATIENTS
15
0\ _._ 1*
1
10
I
t
-A-
Tarrytown, NY). Fibrinogen concentration was measured using a nephelometric centrifugal analyzer (Automated Coagulation Laboratory, Instrumentation Laboratory, Lexington, MA). Ultrafiltered volume was determined from intradialytic weight loss.
Control CAPD
Statistical Analysis
HD
Comparisons between groups were determined by analysis of variance (ANOVA) and Duncan's multiple range test using SAS 17 on an IBM mainframe computer (IBM, Yorktown Heights, NY). Comparisons between values before and after HD (effect of HD) were assessed by paired t test. Linear regression analysis was performed for all individual viscosity measurements, and pre- and post-HD plasma total protein, fibrinogen, albumin, cholesterol, and triglyceride concentrations and HCTs. Multiple linear regression analysis using SAS software was performed for age, HCT, and albumin, globulin, cholesterol, and triglyceride concentration to assess their independent effects in determining viscosity. Differences among groups, and correlations were considered significant at P < 0.05. All numerical results are expressed as the mean ±SEM.
~t
t==~\\j
~
'00 o u
en
:>
-0-
~\~"-~ t 'i~. 0____..
5
.--....°
'---
o 2
133
RESULTS
.10
100
Shear Rate
Comparison of Viscosity in Renal Replacement Therapies
(sec- 1)
Fig 1. Whole blood viscosity at different shear rates in normal control subjects and ESRD patients. "Different from others; tdifferent from HD, P < 0.05, ANOVA. Cp, centipoise.
librium centrifugation, and the 60% transitional density range (T60) over which the middle or transitional 60% of the cell population is located.
Biochemical Methods The HCT offresh blood containing EDTA-2Na (I mg/mL whole blood) as anticoagulant was determined in triplicate by standard microhematocrit tube method, by centrifuging at 13,460 X g for 15 minutes and measuring the proportion of packed RBC volume. Total plasma protein, albumin, cholesterol, and triglyceride concentrations were measured using an Technicon Dax 96 autoanalyzer (Technicon Instrument,
The mean whole blood viscosity was higher in the control subjects than both CAPD patients at the lowest shear rate (2.3 S-l) and HD patients before treatment at shear rates of 2.3 to 115 S-l applied under steady shear flow conditions (Fig 1). There was no significant difference between the mean whole blood viscosity values in HD and CAPD patients. These findings can be at least partially explained by the differences in HCTs between the populations. The CAPD patients had a mean HCT of 0.28 ± 0.01, similar to the mean HCT of the HD patients at the beginning of treatment (0.27 ± 0.03), while the mean HCT was higher in the control subjects (0.40 ± 0.01) (Table 1). To exclude the effect ofHCT on viscosity, we
Table 1. HCTs and Blood Chemistries in Patients With ESRD Controls
HCT Total protein concentration (giL) Albumin concentration (giL) Fibrinogen concentration (giL) Cholesterol (mmoI/L) Triglycerides (mmoIfL)
0.40 63 37 1.6 4.60 1.25
* Different from other groups, ANOVA, P < 0.05.
± ± ± ± ± ±
0.01* 1 2 0.01* 0.29 0.16
HD
0.27 73 38 3.1 3.90 1.49
± ± ± ± ± ±
CAPD
0.03 3* 2 0.4 0.28 0.25
0.28 62 30 4.0 4.55 1.60
± ± ± ± ± ±
0.01 4 2* 0.6 0.36 0.39
FERIANI ET AL
plasma. The final mean HCTs were similar in all three groups (control, 0.69 ± 0.004 [range, 0.67 to 0.71]; CAPD, 0.67 ± 0.01 [range, 0.65 to 0.70]; HD, 0.68 ± 0.01 [range, 0.65 to 0.72]). At these higher, but equal, HCTs, CAPD patients had higher mean blood viscosity compared with the control subjects at the lower shear rates. Both CAPD and HD patients had higher mean viscosity values than control subjects at shear rates of23 and 46 S-I (Fig 3). At the shear rate of 115 S-I, CAPD patients had higher mean blood viscosity, at the 70% HCT, than HD patients or control subjects. Effect of Hemodialysis on Viscosity and RBC Density
o 2
·10
100
After 1 hour of treatment, the mean whole blood viscosity of the HD patients was still lower than that of the control subjects (data not shown). However, the mean HCT of the HD patients after 1 hour of dialysis was higher than the pretreatment value (0.30 ± 0.03 v 0.27 ± 0.03, P < 0.003).
Shear Rate (sec- 1) 120
Fig 2. Plasma viscosity at different shear rates in normal control subjects and ESRD patients. *Different from others; *different from control, P < 0.05 by ANOVA.
studied plasma viscosity in the subjects. The mean plasma viscosity of CAPD patients was significantly higher than that of both HD patients and control subjects at shear rates of 11.5 to 46 S-I, and higher than control subjects at shear rates greater than 115 S-I (Fig 2). However, the mean plasma total protein concentration was highest in HD patients (Table 1). The mean plasma albumin concentration was lower in CAPD patients than HD patients, or normal control subjects. The mean plasma fibrinogen concentration was higher in patients with ESRD treated with either modality compared with controls. There was no significant difference in mean fibrinogen concentration between patients treated with HD and CAPD (Table 1). There were no differences in mean plasma cholesterol or triglyceride concentrations between the three groups (Table 1). To assess the rheological effects of hemoconcentration in ESRD patients, and further control for the effect of HCT, the HCTs of blood from the control subjects and the patients were adjusted to approximately 70% using autologous
90 .........
c..
u .......... >.
."!:::
en
0
u
60
en
:>
- 0 - Control
-.- CAPO
-,.-
A
~~j
HO
'
~~ ~~~*
30
*~I~.
o 2
10
100
Shear Rate (sec- 1) Fig 3. Whole blood viscosity at high HCT at different shear rates in normal control subjects and ESRD patients. *Different from others; *different from control subjects, P < 0.05 by ANOVA.
135
VISCOSITY IN HEMODIALYSIS AND CAPO PATIENTS
15
1*
0
......... 10
c... ()
->.
+-'
'00 0
1\,** .-~:
Pre HD
-&- Post HD
• _______ I
I~
() (/)
:>
-.-
5
I
~.
o .1 0
Hematocrit Total protein concentration (giL) Albumin concentration (gIL) Fibrinogen concentration (gIL)
100
Shear Rate (sec- 1)
Pre·HD
Post·HD
0.27 ± 0.03
0.31 ± 0.03'
73 ± 3
88 ± 4t
38 ± 2
45 ± 2t
3.1 ± 0.4
3.5 ± 0.4
• P < 0.02, pre v post, paired t test. t P < 0.01, pre v post, paired t test.
............... _ I I l ___
* *.---.-----• * • * 2
Table 2. Effect of Hemodialysis on Hematocrit and Blood Chemistries
- 0 - Control
As measured by the phthalate ester technique, the density profile of the RBCs of treated HD patients increased compared with the pretreatment profile. The absolute change in the median density (post-HD D50 - pre-HD D 50) of the RBCs of the treated patients ranged from 0.0005 to
12
Fig 4. Whole blood viscosity at different shear rates in HD patients: effect of HD. *Different from others; ""different from pre-HD, P < 0.05 by ANOVA.
- 0 - Control -&-
Both the mean whole blood and plasma viscosities of the HD patients increased during the first hour of treatment, concomitantly with ultrafiltration. The mean whole blood viscosity of HD patients attained a value similar to the mean normal controls' values at the end of dialysis at almost all shear rates (Fig 4). Whole blood viscosity at high shear rates was higher after treatment compared with values before dialysis in HD patients. The HCT of the HD patients following treatment reached a value of 0.31 ± 0.03, still lower than that of the control subjects, but increased 13.5% compared with the mean pre-HD value (P < 0.02) (Table 2). The posttreatment plasma viscosity in HD patients also increased to a value higher than that of control subjects, but the difference was significant only at high shear rates (Fig 5). The mean plasma viscosity at the end of dialysis was higher than that of both control subjects and the mean pre-HD values, coincident with ultrafiltration.
-
a:
Pre HD
- . - Post HD
8
()
>. ...... '00
o
() (/)
:>
4
o 2
10
100
Shear Rate (sec- 1) Fig 5. Plasma viscosity at different shear rates in HD patients: effect of HD. "Different from others; *different from pre-HD, P < 0.05 by ANOVA.
FERIANI ET AL
136
Table 3. Correlation of HCT and Whole Blood Viscosity in Normal Subjects and Patients With ESRD All subjects (n = 26) Shear rate S-l
r
p
Controls (n = 9) Shear rate S-l
r
p
ESRO patients (n Shear rate S-l
r P HO (n = 9) Shear rate S-l
r P CAPO (n = 8) Shear rate S-l
r
p
=
230 .911 .0001
115 .906 .0001
46 .894 .0001
23 .695 .0001
11.5 .595 .0017
5.75 .456 .022
2.3 .568 .0031
230 .934 .0001
115 .979 .0001
46 .895 .0027
23 .780 .0225
11.5 .776 .0237
5.75 .209 .62
2.3 -.022 .958
230 .862 .0001
115 .845 .0001
46 .796 .0001
23 .519 .033
11 .5 .433 .083
57 .5 .246 .342
2.3 .178 .495
230 .978 .0001
115 .974 .0001
46 .937 .0002
23 .892 .0012
11.5 .701 .035
5.75 .066 .866
2.3 -.016 .967
230 .574 .14
115 .639 .088
46 .667 .071
23 .263 .530
11.5 .269 .519
5.75 .733 .0384
2.3 .776 .024
17)
0.0007 g/mL. Two variable statistics and linear regression analysis showed the absolute change in 0 50 increased linearly as function of ultrafiltration (r = 0.998, P < 0.001). However, the T60 indices did not change compared with the pretreatment values (pre-HO T60, 0.0060 ± 6% v post-HO T60, 0.0069 ± 6.5%, mean ± coefficient of variation), indicating a uniform alteration of the hydration state of the RBC population in these patients.
Relationship of Plasma Proteins and Lipids to Viscosity in ESRD Patients To explain the altered blood rheological characteristics in ESRO patients, we correlated the plasma total protein, albumin, fibrinogen, cholesterol, and triglyceride concentrations with the viscosity data. After treatment, HO patients had significantly higher plasma total protein and albumin levels compared with pretreatment values (Table 2). The increase in fibrinogen levels was not significant, and there was no change in mean plasma cholesterol or triglyceride levels when values before and after treatment were compared. Linear regression analysis was performed for all individual viscosity measurements at the respective shear rates, and the individual values of pre- and post-HO treatment plasma total protein, fibrinogen, albumin, cholesterol, and triglyceride concentrations. The only significant relationships
were between viscosity and total protein and fibrinogen levels at shear rates of 115 and 230 S-1 (total protein: r = 0.59, P < 0.05, and r = 0.80, P < 0.02; fibrinogen: r = 0.64, P < 0.004, and r = 0.72, P < 0.003, respectively). When all subjects were considered, there was a significant correlation of viscosity and HCT at all shear rates studied (Table 3). In both the HO patients and control subjects, HCT and viscosity correlated significantly at shear rates of 11.5 to 230 S-I. When all 17 ESRO patients' values were analyzed, HCT and viscosity showed a similar pattern, correlating at shear rates of 23 to 230 s- I. In contrast, in CAPO patients, these two parameters significantly correlated only at the lowest shear rates (5.75 and 2.3 S-I). There were no significant correlations of viscosity with plasma protein or lipid concentrations in patients treated with HO. In contrast, there was an inverse correlation of viscosity with triglyceride concentration at shear rates of 23 to 230 s- I, and with cholesterol concentration at shear rates of 11.5 and 23 S-I. The relationship between viscosity and fibrinogen in patients treated with CAPO approached significance only at the highest shear rates used in the shear field (I 15 S-I, r= 0.74, andP= 0.056;230 S-I, r= 0.71 and P = 0.074). Multiple linear regression analysis for variables determining viscosity at shear rates of 115 and
VISCOSITY IN HEMODIALYSIS AND CAPO PATIENTS
137
11.5 S-I showed the majority of the variance in viscosity was accounted for by HCT. Only the effect ofHCTwas significant (P < 0.001, r = 0.93; r = 0.90, P < 0.05, respectively). Age did not account for the variance in viscosity in the linear regression model, and was uncorrelated with HCT and viscosity in control subjects and HD patients. In contrast, in CAPD patients, age and viscosity were correlated (r = 0.90, P < 0.003).
increased, despite significantly lower HCT levels after HD (compared with control samples'), again demonstrating the dissociation of HCT and viscosity in HD patients. These findings are similar to those recently reported. 8•9 It is apparent that the increase in viscosity is not solely dependent on increased HCT. An increase in plasma viscosity associated with ultrafiltration was also noted. At the end of treatment, significant increases in total plasma protein and albumin levels in HD patients also undoubtedly contributed to the increase in viscosity. In addition, increases in plasma fibrinogen levels may have also played an important role in modifying viscosity. Since the viscosity of a protein solution depends more on the molecular shape of the protein than its size, the less symmetrical a molecule, the greater effect it exerts on the viscosity of the solution. Fibrinogen, with an axial ratio of 18.75, has far greater effect on the viscosity of plasma than albumin with an axial ratio of 3.95. 19 Therefore, the increase in both fibrinogen and albumin concentration after treatment, and their combined hydrodynamic influence (ie, volume concentration of the proteins, and the hydrodynamic variation of effective molecular volumes) would significantly increase plasma viscosity. In this study, the increased concentration of total plasma protein also contributed significantly to the rheological changes in the patients. The shear thinning of blood as the shear rate increases diminishes its apparent viscosity. However, the physiologic decline ofthe blood shear rate within the microcirculation results in shear-dependent changes in the deformation and aggregation of the RBC. At very low shear rates « lO s- I), the RBC form rouleaux due to the bridging of adjacent cells by fibrinogen, globulins, and other large molecules as a result of the algebraic balance of energies between macromolecular bridging and the electrostatic repulsive energy at the cell surface, the mechanical shear energy, and the energy of cell membrane bending. 2o Increase in the RBC density profile concurrent with treatment suggests an alteration in the hydration state of the cells. The hydration state of the RBC is an important determinant of its volume, electrochemical distribution of sodium, potassium, and chloride, its surface area to volume ratio, and hence its deformability in the microcirculation. The in-
DISCUSSION
Whole blood viscosity was lower in patients with ESRD treated with HD than controls, primarily as a function of anemia. When plasma viscosity was assessed, in order to exclude the effect of HCT, CAPD patients had higher mean viscosity than both control subjects and HD patients. At high shear rates, CAPD patients had higher mean plasma viscosity than the control subjects, suggesting the presence of a circulating factor which influences viscosity in this patient population. Plasma total protein and fibrinogen levels were found to correlate significantly with viscosity in the ESRD patients in this study. Increased levels of fibrinogen in patients treated for ESRD have been previously demonstrated. 4 . 18 Plasma fibrinogen levels were highest in CAPD patients, but the use of ANOV A minimizes the differences between the two patient populations. Plasma viscosity was higher in this group of patients, despite a tendency to low total protein levels, and unequivocal decline in albumin levels. Similar findings were demonstrated when viscosity was assessed at high but equal HCTs, where, at capillary shear rates, CAPD patients manifested significantly higher viscosity. Interestingly, age correlated with viscosity in these patients, but not in the other groups. Perhaps age in these patients is related to a measured or unmeasured circulating substance that affects plasma viscosity. These data demonstrate that viscosity can be dissociated from HCT in ESRD patients, and suggest that a circulating factor in uremic plasma, which may be altered by choice of treatment modality, may be an important modifying determinant of its magnitude. Whole blood and plasma viscosity increased during HD, concomitantly with ultrafiltration. At the end of treatment, whole blood viscosity approached that of control subjects' as the HCT
138
FERIANI ET AL
creased density profile of the RBC also reflects an increase in its internal viscosity,16 and a reduction in its deformability. The reduction in deformability also contributes to the elevation of blood viscosity. These rheological alterations in the RBC and cell-plasma protein interactions may be important pathophysiologically, since such high levels of viscosity may be present within certain segments of the microcirculation where instantaneous local or regional HCTs may reach the levels observed in these experiments. Blood flow through surgically created arteriovenous fistulae has been shown to correlate negatively with whole blood viscosity in patients with ESRD.21 Perhaps such hyperviscosity may mediate clotting of vascular accesses after HD treatments. It is also possible that high levels of viscosity may be associated with the increased prevalence of cerebrovascular and cardiovascular disease seen in patients with ESRD.22 Increased blood viscosity may raise the resistance and therefore the pressure in the extracorporeal circuit during HD. Such rheological disorders related to increased viscosity may be magnified as the mean HCT of the HD population increases, with improved efficiency of hemodialysis and the use of rHuEPO. 9 The role of viscosity in patients with CAPD has not been intensively investigated. Previous work 13 has suggested that w-3 polyunsaturated acid administration decreases the viscosity of whole blood obtained from CAPD patients reconstituted at high HCT. Unfortunately, no comparisons at lower HCTs, typical of the anemia of ESRD, or between patients treated with HD, and normal controls were reported in that study. The reason for different patterns of correlation of HCT and viscosity in CAPD patients compared with HD patients or normal control
subjects, observed in the present study, is unknown. Such differences may be related to differences in the pattern of total cholesterol and cholesterol fraction distribution seen in patients treated with CAPD and HD.2 3 .24 In light of these findings, the inverse correlation of plasma lipid concentration and viscosity in patients treated with CAPD noted in our study is especially noteworthy. The increase in fibrinogen in ESRD patients has been previously documented. 4,18 There may be a decrease in a protein or proteins in CAPD patients that alters fibrinogen metabolism or catabolism secondary to net losses in peritoneal effluents. 25 Hyperfibrinogenemia, noted in patients with chronic renal insufficiency, may predispose to progressive renal disease 26 by increasing fibrin deposition or procoagulant activity in glomeruli. In addition, increased fibrinogen levels have been associated with increased cardiovascular morbidity and mortality,27-29 and may playa role in mediating the prevalence of cardiovascular and cerebrovascular disease in patients with ESRD. The relatively higher fibrinogen levels in CAPD patients compared with HD patients may have important implications regarding patient survival. These issues deserve further investigation. ACKNOWLEDGMENT The authors would like to thank Jorge Rios, MD, and Milton Gusack, MD, and the Rita Gusack Fund for their kind support. We thank Samuel Simmens, PhD, of the Department of Health Care Sciences, for assistance with statistical analysis. The authors appreciate the expert technical assistance of Christopher Gubish, MS, and Toufic Safa, MD, and the assistance of Glenda McArthur, Teena Louise Ballard, and Adriana Flury de Gonzalez in manuscript preparation. We thank the entire nursing staff of the George Washington University Medical Center Ambulatory Dialysis Center, including Joan Watson, RN, and Farimah Azali, RN, for help in collecting samples. Most importantly, we thank our patients for participating in this study.
REFERENCES I. Raine AEG: Hypertension, blood viscosity, and cardiovascular morbidity in renal failure: Implications of erythropoietin therapy. Lancet 1:97-100, 1988 2. Tibblin A, Bergentz S-E, Bjure J, et al: Hematocrit, plasma proteins, plasma volume, and viscosity in early hypertensive disease. Am Heart J 72:165-176, 1966 3. Devereux RB, Drayer JIM, Chien S: Whole blood viscosity as a determinant of cardiac hypertrophy in systemic hypertension. Am J Cardiol 54:592-595, 1984 4. Shusterman N, Kimmel P, Kiechle FL, et a1: Factors influencing erythrocyte sedimentation in patients with
chronic renal failure. Arch Intern Med 145:1796-1799, 1985 5. Benis AM, Chien S, Usami S, Jan K-M: Inertial pressure losses in the perfused hind limb: A reinterpretation of the results of Whittacker and Winton. J Appl PhysioI34:383-389, 1973 6. Neff MS, Kim KE, Persoff M, et al: Hemodynamics of uremic anemia. Circulation 43:876-883, 1971 7. Capelli JP, Kasparian H: Cardiac work demands and left ventricular function in end stage renal disease. Ann Intern Med 86:261-267, 1977
VISCOSITY IN HEMODIALYSIS AND CAPO PATIENTS
139
8. Wink 1, Vaziri ND, Barker S, et al: The effect of hemodialysis on whole blood, plasma and erythrocyte viscosity. Int 1 Artif Organs 11 :340-342, 1988 9. Vaziri ND, Ritchie C, Brown P, et al: Effect of erythropoietin administration on blood and plasma viscosity in hemodialysis patients. ASAIO Trans 35:505-508, 1989 10. Schaefer RM, Leschke M, Strauer BE, et al: Blood rheology and hypertension in hemodialysis patients treated with erythropoietin. Am 1 Nephrol 8:449-53, 1988 11. Steffen HM, Brunner R, Muller R, et al: Peripheral hemodynamics, blood viscosity, and the renin-angiotensin system in hemodialysis patients under therapy with recombinant human erythropoietin. Contrib Nephrol 76:292-298, 1989 12. Brunner R, Steffen HM, Pollok M, et al: Blood rheology in hemodialysis patients treated with recombinant human erythropoietin. Contrib Nephrol 76:306-312, 1989 13. van Acker BA, Bilo Hl, Popp-Snijders C: The effect of fish oil on lipid profile and viscosity of erythrocyte suspensions in CAPD patients. Nephrol Dial Transplant 2:557-561, 1987 14. Wells RE, Denton R, Merrill EW: Measurement of viscosity of biologic fluids by cone plate viscometer. 1 Lab Clin Med 57:646-656, 1961 15. ReindorfCA, Kurantsin-Mills 1, Allotey lB, et al: Perfluorocarbon compounds: Effect on the rheological properties of sickle erythrocytes in. vitro. Am 1 Hematol 19:229-236, 1985 16. Kurantsin-Mills 1, lacobs HM, Lessin LS: Assessment of the hydration state of sickle cells by phthalate ester density distribution. 1 Lab Clin Med 109:486-494, 1987 17. Ray AA: SAS User's Guide: Statistics. Cary, NC, SAS Institute, 1982, pp 113-137 18. Panicucci F, Sagripanti A, Pinori E, et al: Comprehensive study ofhaemostasis in chronic uraemia. Nephron 33:58, 1983
19. Bull HB: Viscosity and the flow of liquids, in Bull HB (ed): An Introduction to Physical Biochemistry. Philadelphia, PA, Davis, 1971, pp 261-275 20. Chien S: Aggregation of red blood cells: an electrochemical and colloid chemical problem, in Blank M (ed): Bioelectrochemistry: Ions, Surface and Membranes. Washington, DC, American Chemical Society, 1980, pp 3-38 21. Reilly DT, Wood FM, Bell RF: Arteriovenous fistulas for dialysis: Blood flow, viscosity, and long-term patency. World 1 Surg 6:628-633, 1982 22. Lindner A, Charra B, Sherrard Dl, et al: Accelerated atherosclerosis and prolonged maintenance hemodialysis. N Eng! 1 Med 290:679-701,1974 23. Avram MM, Fein PA, Antignani A, et al: Cholesterol and lipid disturbances in renal disease: The natural history of uremic dyslipidemia and the impact of hemodialysis and continuous ambulatory peritoneal dialysis. Am 1 Med 87:555560, 1989 24. Appel G: Lipid abnormalities in renal disease. Kidney Int 39: 169-183, 1991 25. Lai KN, Yin lA, Yven PMP, et al: Protein C, protein S, and antithrombin III levels in patients on continuous ambulatory peritoneal dialysis, and hemodialysis. Nephron 56: 271-276, 1990 26. Klahr S, Schreiner G, Ichikawa I: The progression of renal disease. N Eng! 1 Med 318:1657-1666,1988 27. Kannel WB, Wo1fPA, Castelli WP, et al: Fibrinogen and risk of cardiovascular disease. lAMA 258: 1183-1186, 1987 28. Wilhelmsen L, Svardsudd K, Korsan-Bengtsen K, et al: Fibrinogen as a risk factor for stroke and myocardial infraction. N Eng! 1 Med 311:501-505, 1984 29. Di Minno G, Mancini M: Measuring plasma fibrinogen to predict stroke and myocardial infraction. Arteriosclerosis 10:1-7,1990
W
HOLE BLOOD viscosity is a function of plasma viscosity and hematocrit (HCT) levels. I Increased blood viscosity has been associated with increased peripheral vascular resistance, hypertension, and cardiac hypertrophy. I Whole blood viscosity is increased in patient populations with renal disease and hypertension. 2-4 The mechanisms underlying the hemodynamic changes in patients with renal disease are not well understood. However, a number of studies have demonstrated a direct relation between HCT and vascular resistance. 3,5-7 Other factors such as increased levels of fibrinogen may be associated with hyperviscosity in patients with chronic renal insufficiency, and patients with end-stage renal disease (ESRD) treated with hemodialysis (HD).4 Treatment of patients with ESRD by HD, intermittent peritoneal dialysis, or continuous ambulatory peritoneal dialysis (CAPD) results in the reduction of plasma volume, improvement of anemia, and clearance of uremic toxins. Although anemia is a problem
in the ESRD population, hemoconcentration may occur in patients, coincident with net ultrafiltration. Whole blood and plasma viscosity in patients with ESRD treated with HD has been shown to increase after treatment, associated with weight loss and increase in HCT level. 8 While viscosity is a function of HCT, and its alterations might From the Department o/Nephrology, St. Bortolo Hospital, Vicenza, Italy; and the Department 0/ Medicine, George Washington University Medical Center, Washington, DC. Supported in part/rom/unds provided by the National Institutes 0/ Health (Grant No. HL-38521 to J K-M) and the Department 0/ Medicine, George Washington University Medical Center. Presented in part at the 23rd Annual meeting a/the American Society a/Nephrology, Washington, DC, December 1990. Address reprint requests to Paul L. Kimmel, MD, Division 0/R enal Diseases and Hypertension, Department a/Medicine, George Washington University, 2150 Pennsylvania Ave, NW, Washington, DC 20037. © 1992 by the National Kidney Foundation, Inc. 0272-6386/92/1902-0004$3.00/0
American Journal of Kidney Diseases, Vol XIX, No 2 (February), 1992: pp 131-139
131
FERIANI ET AL
132
contribute to increased peripheral vascular resistance, its specific role, if any, in affecting mortality in such patients is unknown. The recent successful use of recombinant DNA human erythropoietin (rHuEPO) to treat the anemia of ESRD has raised concern about the development or worsening of hypertension in a population of treated ESRD patients because of the increased HCT and the associated elevated blood viscosity.· Several studies9 - •• demonstrated increases in whole blood viscosity over time in HD patients treated with rHuEPO; however, these findings have been disputed. 12 Vaziri et al9 showed an increase in whole blood viscosity in rHuEPO-treated HD patients when postdialysis treatment values were compared with baseline values. Finally, few studies of viscosity in patients with ESRD treated with CAPD have been performedY Comparisons between viscosity measurements before and after HD treatment with values in patients treated with CAPD have not been made. To assess the rheological factors associated with ESRD and its treatment, and to compare the effects of different renal replacement therapies, we investigated the effect of dialysis on whole blood and plasma viscosity in patients treated with CAPD and HD under steady shear flow conditions, and compared the data with those of normal healthy subjects. METHODS
Patients Nine patients with ESRD treated with HD, aged 36 to 79 years (mean, 60 ± 5), volunteered for the study. Three patients had diabetic nephropathy, three had hypertensive nephropathy, and one each had obstructive nephropathy, chronic glomerulonephritis, and unknown renal disease. AU patients were either dialyzed with high-efficiency or high-flux dialyzers, three times weekly. Weight was measured before and after each treatment. All patients were clinically stable at the time of investigation and had been dialyzed for at least I year. Whole blood samples were drawn from the arterial side of the circuit just after the beginning of dialysis, and just before the end of dialysis_Blood was obtained in all cases by venipuncture, using disodium ethylenediamine-tetraacetate (EDTA-2Na) I mg! mL whole blood as an anticoagulant. Eight patients with ESRD treated with CAPD, aged 23 to 62 years (mean, 36 ± 5), were studied simultaneously. Two patients had diabetic nephropathy, three had hypertensive nephropathy, two had unknown renal disease, and one had chronic renal failure complicating horseshoe kidney. Blood samples were obtained at the time of monthly routine visits. Nine healthy subjects, aged 30 to 60 years (mean, 39 ± 4), served as a control population.
All patients gave their informed consent before participating. The study was approved by the George Washington University Medical Center Committee on Human Research.
Determination of Viscosity Measurements of viscosity were made on I-ml aliquots of normal human autologous plasma (as controls) and l-mL aliquots of whole blood of the patients. The cell preparations were maintained at a constant temperature of 37°C. All viscosity measurements were performed within 2 hours of blood collection, in triplicate for each sample. Intrasample coefficient of variation was approximately 5%. To assess the effect of red blood cell (RBC) concentration on viscosity, the whole blood was also adjusted to approximately 70% HCT for the control and patient samples using autologous plasma, and used to measure the viscosity. This required centrifuged separation of the blood for 15 minutes at 2,000 X g, 25°C, and resuspension of the cells in the autologous plasma only. Aliquots (I ml) of the resuspended high HCT blood of control samples or 1ml aliquots of samples of ESRD patients were used for the viscosity measurements. Viscosity was measured in a WellsBrookfield model LVT cone-plate microviscometer (Brookfield Engineering Labs, Stoughton, MA) at a constant temperature of 37°C. Shear rate, calculated from rotation speed and the geometry of the cone and plate, varied from 230 to 2.3 S- I. Shear stress was calculated from the torque (T in dyne centimeters) acting on the cone of the viscometer. 14.15 Apparent viscosity (1/, hereafter referred to as viscosity) is defined as the ratio shear stress to shear rate in dyne· s/cm 2 or poise: viscosity = shear stress (dynes/cm 2)/shear rate (S-I). The result multiplied by 100 yields the unit centipoise (cp).
Determination of RBC Density Profiles The density distribution profile of the RBCs was determined before and after treatment in HD patients, using the phthalate ester density technique. The details of the technique have been described elsewhere. 16 Briefly, the density distribution profiles of the various cell suspensions were analyzed in phthalate esters prepared from 20 mixtures of di-n-butyl phthalate (specific gravity, 1.042) and dimethyl phthalate (specific gravity, \.189) (Fisher Scientific, Pittsburgh, PAl with final specific gravity of the mixtures between 1.062 and 1.138 in increments of 0.004. Duplicate 60-IlL samples of the blood preparations were pipetted into microcapillary tubes (75-mm long, 1.15mm internal diameter) containing 10 ILL of each of the phthalate ester mixtures. The end of the microcapillary tube in contact with the phthalate was sealed by flaming. The microcapillary tubes were then centrifuged at 13,460 X g, for 15 minutes, during which time the temperature equilibrated at 36.5°C (IEC MC Microhematocrit Centrifuge, Needham Heights, MA). The distribution of RBCs was analyzed by their differential floatation on the esters. The percentage of RBCs below the ester after the equilibrium centrifugation was related to the ester density using a computer program (PEDDS; Division of Hematology-Oncology, George Washington University Medical Center). The density profile curves generated from the data were analyzed for numerical indices that quantitatively described the properties of the cell population. These indices are the median density (D50), the density of the phthalate ester at which 50% of the cells are located after equi-
VISCOSITY IN HEMODIALYSIS AND CAPD PATIENTS
15
0\ _._ 1*
1
10
I
t
-A-
Tarrytown, NY). Fibrinogen concentration was measured using a nephelometric centrifugal analyzer (Automated Coagulation Laboratory, Instrumentation Laboratory, Lexington, MA). Ultrafiltered volume was determined from intradialytic weight loss.
Control CAPD
Statistical Analysis
HD
Comparisons between groups were determined by analysis of variance (ANOVA) and Duncan's multiple range test using SAS 17 on an IBM mainframe computer (IBM, Yorktown Heights, NY). Comparisons between values before and after HD (effect of HD) were assessed by paired t test. Linear regression analysis was performed for all individual viscosity measurements, and pre- and post-HD plasma total protein, fibrinogen, albumin, cholesterol, and triglyceride concentrations and HCTs. Multiple linear regression analysis using SAS software was performed for age, HCT, and albumin, globulin, cholesterol, and triglyceride concentration to assess their independent effects in determining viscosity. Differences among groups, and correlations were considered significant at P < 0.05. All numerical results are expressed as the mean ±SEM.
~t
t==~\\j
~
'00 o u
en
:>
-0-
~\~"-~ t 'i~. 0____..
5
.--....°
'---
o 2
133
RESULTS
.10
100
Shear Rate
Comparison of Viscosity in Renal Replacement Therapies
(sec- 1)
Fig 1. Whole blood viscosity at different shear rates in normal control subjects and ESRD patients. "Different from others; tdifferent from HD, P < 0.05, ANOVA. Cp, centipoise.
librium centrifugation, and the 60% transitional density range (T60) over which the middle or transitional 60% of the cell population is located.
Biochemical Methods The HCT offresh blood containing EDTA-2Na (I mg/mL whole blood) as anticoagulant was determined in triplicate by standard microhematocrit tube method, by centrifuging at 13,460 X g for 15 minutes and measuring the proportion of packed RBC volume. Total plasma protein, albumin, cholesterol, and triglyceride concentrations were measured using an Technicon Dax 96 autoanalyzer (Technicon Instrument,
The mean whole blood viscosity was higher in the control subjects than both CAPD patients at the lowest shear rate (2.3 S-l) and HD patients before treatment at shear rates of 2.3 to 115 S-l applied under steady shear flow conditions (Fig 1). There was no significant difference between the mean whole blood viscosity values in HD and CAPD patients. These findings can be at least partially explained by the differences in HCTs between the populations. The CAPD patients had a mean HCT of 0.28 ± 0.01, similar to the mean HCT of the HD patients at the beginning of treatment (0.27 ± 0.03), while the mean HCT was higher in the control subjects (0.40 ± 0.01) (Table 1). To exclude the effect ofHCT on viscosity, we
Table 1. HCTs and Blood Chemistries in Patients With ESRD Controls
HCT Total protein concentration (giL) Albumin concentration (giL) Fibrinogen concentration (giL) Cholesterol (mmoI/L) Triglycerides (mmoIfL)
0.40 63 37 1.6 4.60 1.25
* Different from other groups, ANOVA, P < 0.05.
± ± ± ± ± ±
0.01* 1 2 0.01* 0.29 0.16
HD
0.27 73 38 3.1 3.90 1.49
± ± ± ± ± ±
CAPD
0.03 3* 2 0.4 0.28 0.25
0.28 62 30 4.0 4.55 1.60
± ± ± ± ± ±
0.01 4 2* 0.6 0.36 0.39
FERIANI ET AL
plasma. The final mean HCTs were similar in all three groups (control, 0.69 ± 0.004 [range, 0.67 to 0.71]; CAPD, 0.67 ± 0.01 [range, 0.65 to 0.70]; HD, 0.68 ± 0.01 [range, 0.65 to 0.72]). At these higher, but equal, HCTs, CAPD patients had higher mean blood viscosity compared with the control subjects at the lower shear rates. Both CAPD and HD patients had higher mean viscosity values than control subjects at shear rates of23 and 46 S-I (Fig 3). At the shear rate of 115 S-I, CAPD patients had higher mean blood viscosity, at the 70% HCT, than HD patients or control subjects. Effect of Hemodialysis on Viscosity and RBC Density
o 2
·10
100
After 1 hour of treatment, the mean whole blood viscosity of the HD patients was still lower than that of the control subjects (data not shown). However, the mean HCT of the HD patients after 1 hour of dialysis was higher than the pretreatment value (0.30 ± 0.03 v 0.27 ± 0.03, P < 0.003).
Shear Rate (sec- 1) 120
Fig 2. Plasma viscosity at different shear rates in normal control subjects and ESRD patients. *Different from others; *different from control, P < 0.05 by ANOVA.
studied plasma viscosity in the subjects. The mean plasma viscosity of CAPD patients was significantly higher than that of both HD patients and control subjects at shear rates of 11.5 to 46 S-I, and higher than control subjects at shear rates greater than 115 S-I (Fig 2). However, the mean plasma total protein concentration was highest in HD patients (Table 1). The mean plasma albumin concentration was lower in CAPD patients than HD patients, or normal control subjects. The mean plasma fibrinogen concentration was higher in patients with ESRD treated with either modality compared with controls. There was no significant difference in mean fibrinogen concentration between patients treated with HD and CAPD (Table 1). There were no differences in mean plasma cholesterol or triglyceride concentrations between the three groups (Table 1). To assess the rheological effects of hemoconcentration in ESRD patients, and further control for the effect of HCT, the HCTs of blood from the control subjects and the patients were adjusted to approximately 70% using autologous
90 .........
c..
u .......... >.
."!:::
en
0
u
60
en
:>
- 0 - Control
-.- CAPO
-,.-
A
~~j
HO
'
~~ ~~~*
30
*~I~.
o 2
10
100
Shear Rate (sec- 1) Fig 3. Whole blood viscosity at high HCT at different shear rates in normal control subjects and ESRD patients. *Different from others; *different from control subjects, P < 0.05 by ANOVA.
135
VISCOSITY IN HEMODIALYSIS AND CAPO PATIENTS
15
1*
0
......... 10
c... ()
->.
+-'
'00 0
1\,** .-~:
Pre HD
-&- Post HD
• _______ I
I~
() (/)
:>
-.-
5
I
~.
o .1 0
Hematocrit Total protein concentration (giL) Albumin concentration (gIL) Fibrinogen concentration (gIL)
100
Shear Rate (sec- 1)
Pre·HD
Post·HD
0.27 ± 0.03
0.31 ± 0.03'
73 ± 3
88 ± 4t
38 ± 2
45 ± 2t
3.1 ± 0.4
3.5 ± 0.4
• P < 0.02, pre v post, paired t test. t P < 0.01, pre v post, paired t test.
............... _ I I l ___
* *.---.-----• * • * 2
Table 2. Effect of Hemodialysis on Hematocrit and Blood Chemistries
- 0 - Control
As measured by the phthalate ester technique, the density profile of the RBCs of treated HD patients increased compared with the pretreatment profile. The absolute change in the median density (post-HD D50 - pre-HD D 50) of the RBCs of the treated patients ranged from 0.0005 to
12
Fig 4. Whole blood viscosity at different shear rates in HD patients: effect of HD. *Different from others; ""different from pre-HD, P < 0.05 by ANOVA.
- 0 - Control -&-
Both the mean whole blood and plasma viscosities of the HD patients increased during the first hour of treatment, concomitantly with ultrafiltration. The mean whole blood viscosity of HD patients attained a value similar to the mean normal controls' values at the end of dialysis at almost all shear rates (Fig 4). Whole blood viscosity at high shear rates was higher after treatment compared with values before dialysis in HD patients. The HCT of the HD patients following treatment reached a value of 0.31 ± 0.03, still lower than that of the control subjects, but increased 13.5% compared with the mean pre-HD value (P < 0.02) (Table 2). The posttreatment plasma viscosity in HD patients also increased to a value higher than that of control subjects, but the difference was significant only at high shear rates (Fig 5). The mean plasma viscosity at the end of dialysis was higher than that of both control subjects and the mean pre-HD values, coincident with ultrafiltration.
-
a:
Pre HD
- . - Post HD
8
()
>. ...... '00
o
() (/)
:>
4
o 2
10
100
Shear Rate (sec- 1) Fig 5. Plasma viscosity at different shear rates in HD patients: effect of HD. "Different from others; *different from pre-HD, P < 0.05 by ANOVA.
FERIANI ET AL
136
Table 3. Correlation of HCT and Whole Blood Viscosity in Normal Subjects and Patients With ESRD All subjects (n = 26) Shear rate S-l
r
p
Controls (n = 9) Shear rate S-l
r
p
ESRO patients (n Shear rate S-l
r P HO (n = 9) Shear rate S-l
r P CAPO (n = 8) Shear rate S-l
r
p
=
230 .911 .0001
115 .906 .0001
46 .894 .0001
23 .695 .0001
11.5 .595 .0017
5.75 .456 .022
2.3 .568 .0031
230 .934 .0001
115 .979 .0001
46 .895 .0027
23 .780 .0225
11.5 .776 .0237
5.75 .209 .62
2.3 -.022 .958
230 .862 .0001
115 .845 .0001
46 .796 .0001
23 .519 .033
11 .5 .433 .083
57 .5 .246 .342
2.3 .178 .495
230 .978 .0001
115 .974 .0001
46 .937 .0002
23 .892 .0012
11.5 .701 .035
5.75 .066 .866
2.3 -.016 .967
230 .574 .14
115 .639 .088
46 .667 .071
23 .263 .530
11.5 .269 .519
5.75 .733 .0384
2.3 .776 .024
17)
0.0007 g/mL. Two variable statistics and linear regression analysis showed the absolute change in 0 50 increased linearly as function of ultrafiltration (r = 0.998, P < 0.001). However, the T60 indices did not change compared with the pretreatment values (pre-HO T60, 0.0060 ± 6% v post-HO T60, 0.0069 ± 6.5%, mean ± coefficient of variation), indicating a uniform alteration of the hydration state of the RBC population in these patients.
Relationship of Plasma Proteins and Lipids to Viscosity in ESRD Patients To explain the altered blood rheological characteristics in ESRO patients, we correlated the plasma total protein, albumin, fibrinogen, cholesterol, and triglyceride concentrations with the viscosity data. After treatment, HO patients had significantly higher plasma total protein and albumin levels compared with pretreatment values (Table 2). The increase in fibrinogen levels was not significant, and there was no change in mean plasma cholesterol or triglyceride levels when values before and after treatment were compared. Linear regression analysis was performed for all individual viscosity measurements at the respective shear rates, and the individual values of pre- and post-HO treatment plasma total protein, fibrinogen, albumin, cholesterol, and triglyceride concentrations. The only significant relationships
were between viscosity and total protein and fibrinogen levels at shear rates of 115 and 230 S-1 (total protein: r = 0.59, P < 0.05, and r = 0.80, P < 0.02; fibrinogen: r = 0.64, P < 0.004, and r = 0.72, P < 0.003, respectively). When all subjects were considered, there was a significant correlation of viscosity and HCT at all shear rates studied (Table 3). In both the HO patients and control subjects, HCT and viscosity correlated significantly at shear rates of 11.5 to 230 S-I. When all 17 ESRO patients' values were analyzed, HCT and viscosity showed a similar pattern, correlating at shear rates of 23 to 230 s- I. In contrast, in CAPO patients, these two parameters significantly correlated only at the lowest shear rates (5.75 and 2.3 S-I). There were no significant correlations of viscosity with plasma protein or lipid concentrations in patients treated with HO. In contrast, there was an inverse correlation of viscosity with triglyceride concentration at shear rates of 23 to 230 s- I, and with cholesterol concentration at shear rates of 11.5 and 23 S-I. The relationship between viscosity and fibrinogen in patients treated with CAPO approached significance only at the highest shear rates used in the shear field (I 15 S-I, r= 0.74, andP= 0.056;230 S-I, r= 0.71 and P = 0.074). Multiple linear regression analysis for variables determining viscosity at shear rates of 115 and
VISCOSITY IN HEMODIALYSIS AND CAPO PATIENTS
137
11.5 S-I showed the majority of the variance in viscosity was accounted for by HCT. Only the effect ofHCTwas significant (P < 0.001, r = 0.93; r = 0.90, P < 0.05, respectively). Age did not account for the variance in viscosity in the linear regression model, and was uncorrelated with HCT and viscosity in control subjects and HD patients. In contrast, in CAPD patients, age and viscosity were correlated (r = 0.90, P < 0.003).
increased, despite significantly lower HCT levels after HD (compared with control samples'), again demonstrating the dissociation of HCT and viscosity in HD patients. These findings are similar to those recently reported. 8•9 It is apparent that the increase in viscosity is not solely dependent on increased HCT. An increase in plasma viscosity associated with ultrafiltration was also noted. At the end of treatment, significant increases in total plasma protein and albumin levels in HD patients also undoubtedly contributed to the increase in viscosity. In addition, increases in plasma fibrinogen levels may have also played an important role in modifying viscosity. Since the viscosity of a protein solution depends more on the molecular shape of the protein than its size, the less symmetrical a molecule, the greater effect it exerts on the viscosity of the solution. Fibrinogen, with an axial ratio of 18.75, has far greater effect on the viscosity of plasma than albumin with an axial ratio of 3.95. 19 Therefore, the increase in both fibrinogen and albumin concentration after treatment, and their combined hydrodynamic influence (ie, volume concentration of the proteins, and the hydrodynamic variation of effective molecular volumes) would significantly increase plasma viscosity. In this study, the increased concentration of total plasma protein also contributed significantly to the rheological changes in the patients. The shear thinning of blood as the shear rate increases diminishes its apparent viscosity. However, the physiologic decline ofthe blood shear rate within the microcirculation results in shear-dependent changes in the deformation and aggregation of the RBC. At very low shear rates « lO s- I), the RBC form rouleaux due to the bridging of adjacent cells by fibrinogen, globulins, and other large molecules as a result of the algebraic balance of energies between macromolecular bridging and the electrostatic repulsive energy at the cell surface, the mechanical shear energy, and the energy of cell membrane bending. 2o Increase in the RBC density profile concurrent with treatment suggests an alteration in the hydration state of the cells. The hydration state of the RBC is an important determinant of its volume, electrochemical distribution of sodium, potassium, and chloride, its surface area to volume ratio, and hence its deformability in the microcirculation. The in-
DISCUSSION
Whole blood viscosity was lower in patients with ESRD treated with HD than controls, primarily as a function of anemia. When plasma viscosity was assessed, in order to exclude the effect of HCT, CAPD patients had higher mean viscosity than both control subjects and HD patients. At high shear rates, CAPD patients had higher mean plasma viscosity than the control subjects, suggesting the presence of a circulating factor which influences viscosity in this patient population. Plasma total protein and fibrinogen levels were found to correlate significantly with viscosity in the ESRD patients in this study. Increased levels of fibrinogen in patients treated for ESRD have been previously demonstrated. 4 . 18 Plasma fibrinogen levels were highest in CAPD patients, but the use of ANOV A minimizes the differences between the two patient populations. Plasma viscosity was higher in this group of patients, despite a tendency to low total protein levels, and unequivocal decline in albumin levels. Similar findings were demonstrated when viscosity was assessed at high but equal HCTs, where, at capillary shear rates, CAPD patients manifested significantly higher viscosity. Interestingly, age correlated with viscosity in these patients, but not in the other groups. Perhaps age in these patients is related to a measured or unmeasured circulating substance that affects plasma viscosity. These data demonstrate that viscosity can be dissociated from HCT in ESRD patients, and suggest that a circulating factor in uremic plasma, which may be altered by choice of treatment modality, may be an important modifying determinant of its magnitude. Whole blood and plasma viscosity increased during HD, concomitantly with ultrafiltration. At the end of treatment, whole blood viscosity approached that of control subjects' as the HCT
138
FERIANI ET AL
creased density profile of the RBC also reflects an increase in its internal viscosity,16 and a reduction in its deformability. The reduction in deformability also contributes to the elevation of blood viscosity. These rheological alterations in the RBC and cell-plasma protein interactions may be important pathophysiologically, since such high levels of viscosity may be present within certain segments of the microcirculation where instantaneous local or regional HCTs may reach the levels observed in these experiments. Blood flow through surgically created arteriovenous fistulae has been shown to correlate negatively with whole blood viscosity in patients with ESRD.21 Perhaps such hyperviscosity may mediate clotting of vascular accesses after HD treatments. It is also possible that high levels of viscosity may be associated with the increased prevalence of cerebrovascular and cardiovascular disease seen in patients with ESRD.22 Increased blood viscosity may raise the resistance and therefore the pressure in the extracorporeal circuit during HD. Such rheological disorders related to increased viscosity may be magnified as the mean HCT of the HD population increases, with improved efficiency of hemodialysis and the use of rHuEPO. 9 The role of viscosity in patients with CAPD has not been intensively investigated. Previous work 13 has suggested that w-3 polyunsaturated acid administration decreases the viscosity of whole blood obtained from CAPD patients reconstituted at high HCT. Unfortunately, no comparisons at lower HCTs, typical of the anemia of ESRD, or between patients treated with HD, and normal controls were reported in that study. The reason for different patterns of correlation of HCT and viscosity in CAPD patients compared with HD patients or normal control
subjects, observed in the present study, is unknown. Such differences may be related to differences in the pattern of total cholesterol and cholesterol fraction distribution seen in patients treated with CAPD and HD.2 3 .24 In light of these findings, the inverse correlation of plasma lipid concentration and viscosity in patients treated with CAPD noted in our study is especially noteworthy. The increase in fibrinogen in ESRD patients has been previously documented. 4,18 There may be a decrease in a protein or proteins in CAPD patients that alters fibrinogen metabolism or catabolism secondary to net losses in peritoneal effluents. 25 Hyperfibrinogenemia, noted in patients with chronic renal insufficiency, may predispose to progressive renal disease 26 by increasing fibrin deposition or procoagulant activity in glomeruli. In addition, increased fibrinogen levels have been associated with increased cardiovascular morbidity and mortality,27-29 and may playa role in mediating the prevalence of cardiovascular and cerebrovascular disease in patients with ESRD. The relatively higher fibrinogen levels in CAPD patients compared with HD patients may have important implications regarding patient survival. These issues deserve further investigation. ACKNOWLEDGMENT The authors would like to thank Jorge Rios, MD, and Milton Gusack, MD, and the Rita Gusack Fund for their kind support. We thank Samuel Simmens, PhD, of the Department of Health Care Sciences, for assistance with statistical analysis. The authors appreciate the expert technical assistance of Christopher Gubish, MS, and Toufic Safa, MD, and the assistance of Glenda McArthur, Teena Louise Ballard, and Adriana Flury de Gonzalez in manuscript preparation. We thank the entire nursing staff of the George Washington University Medical Center Ambulatory Dialysis Center, including Joan Watson, RN, and Farimah Azali, RN, for help in collecting samples. Most importantly, we thank our patients for participating in this study.
REFERENCES I. Raine AEG: Hypertension, blood viscosity, and cardiovascular morbidity in renal failure: Implications of erythropoietin therapy. Lancet 1:97-100, 1988 2. Tibblin A, Bergentz S-E, Bjure J, et al: Hematocrit, plasma proteins, plasma volume, and viscosity in early hypertensive disease. Am Heart J 72:165-176, 1966 3. Devereux RB, Drayer JIM, Chien S: Whole blood viscosity as a determinant of cardiac hypertrophy in systemic hypertension. Am J Cardiol 54:592-595, 1984 4. Shusterman N, Kimmel P, Kiechle FL, et a1: Factors influencing erythrocyte sedimentation in patients with
chronic renal failure. Arch Intern Med 145:1796-1799, 1985 5. Benis AM, Chien S, Usami S, Jan K-M: Inertial pressure losses in the perfused hind limb: A reinterpretation of the results of Whittacker and Winton. J Appl PhysioI34:383-389, 1973 6. Neff MS, Kim KE, Persoff M, et al: Hemodynamics of uremic anemia. Circulation 43:876-883, 1971 7. Capelli JP, Kasparian H: Cardiac work demands and left ventricular function in end stage renal disease. Ann Intern Med 86:261-267, 1977
VISCOSITY IN HEMODIALYSIS AND CAPO PATIENTS
139
8. Wink 1, Vaziri ND, Barker S, et al: The effect of hemodialysis on whole blood, plasma and erythrocyte viscosity. Int 1 Artif Organs 11 :340-342, 1988 9. Vaziri ND, Ritchie C, Brown P, et al: Effect of erythropoietin administration on blood and plasma viscosity in hemodialysis patients. ASAIO Trans 35:505-508, 1989 10. Schaefer RM, Leschke M, Strauer BE, et al: Blood rheology and hypertension in hemodialysis patients treated with erythropoietin. Am 1 Nephrol 8:449-53, 1988 11. Steffen HM, Brunner R, Muller R, et al: Peripheral hemodynamics, blood viscosity, and the renin-angiotensin system in hemodialysis patients under therapy with recombinant human erythropoietin. Contrib Nephrol 76:292-298, 1989 12. Brunner R, Steffen HM, Pollok M, et al: Blood rheology in hemodialysis patients treated with recombinant human erythropoietin. Contrib Nephrol 76:306-312, 1989 13. van Acker BA, Bilo Hl, Popp-Snijders C: The effect of fish oil on lipid profile and viscosity of erythrocyte suspensions in CAPD patients. Nephrol Dial Transplant 2:557-561, 1987 14. Wells RE, Denton R, Merrill EW: Measurement of viscosity of biologic fluids by cone plate viscometer. 1 Lab Clin Med 57:646-656, 1961 15. ReindorfCA, Kurantsin-Mills 1, Allotey lB, et al: Perfluorocarbon compounds: Effect on the rheological properties of sickle erythrocytes in. vitro. Am 1 Hematol 19:229-236, 1985 16. Kurantsin-Mills 1, lacobs HM, Lessin LS: Assessment of the hydration state of sickle cells by phthalate ester density distribution. 1 Lab Clin Med 109:486-494, 1987 17. Ray AA: SAS User's Guide: Statistics. Cary, NC, SAS Institute, 1982, pp 113-137 18. Panicucci F, Sagripanti A, Pinori E, et al: Comprehensive study ofhaemostasis in chronic uraemia. Nephron 33:58, 1983
19. Bull HB: Viscosity and the flow of liquids, in Bull HB (ed): An Introduction to Physical Biochemistry. Philadelphia, PA, Davis, 1971, pp 261-275 20. Chien S: Aggregation of red blood cells: an electrochemical and colloid chemical problem, in Blank M (ed): Bioelectrochemistry: Ions, Surface and Membranes. Washington, DC, American Chemical Society, 1980, pp 3-38 21. Reilly DT, Wood FM, Bell RF: Arteriovenous fistulas for dialysis: Blood flow, viscosity, and long-term patency. World 1 Surg 6:628-633, 1982 22. Lindner A, Charra B, Sherrard Dl, et al: Accelerated atherosclerosis and prolonged maintenance hemodialysis. N Eng! 1 Med 290:679-701,1974 23. Avram MM, Fein PA, Antignani A, et al: Cholesterol and lipid disturbances in renal disease: The natural history of uremic dyslipidemia and the impact of hemodialysis and continuous ambulatory peritoneal dialysis. Am 1 Med 87:555560, 1989 24. Appel G: Lipid abnormalities in renal disease. Kidney Int 39: 169-183, 1991 25. Lai KN, Yin lA, Yven PMP, et al: Protein C, protein S, and antithrombin III levels in patients on continuous ambulatory peritoneal dialysis, and hemodialysis. Nephron 56: 271-276, 1990 26. Klahr S, Schreiner G, Ichikawa I: The progression of renal disease. N Eng! 1 Med 318:1657-1666,1988 27. Kannel WB, Wo1fPA, Castelli WP, et al: Fibrinogen and risk of cardiovascular disease. lAMA 258: 1183-1186, 1987 28. Wilhelmsen L, Svardsudd K, Korsan-Bengtsen K, et al: Fibrinogen as a risk factor for stroke and myocardial infraction. N Eng! 1 Med 311:501-505, 1984 29. Di Minno G, Mancini M: Measuring plasma fibrinogen to predict stroke and myocardial infraction. Arteriosclerosis 10:1-7,1990