A bridge from bioimpedance spectroscopy to 50 kHz bioimpedance analysis: application to total body water measurements.

This paper presents a method for extrapolating the total body water (TBW) resistance R(t50) from the resistance measured at 50 kHz (R(50)). A DXA examination and impedance measurements were carried out in a 1st group of 57 healthy volunteers with a Xitron 4200 multifrequency impedancemeter, in order to determine their values of R(t50) by comparison with resistances extrapolated at an infinite frequency by the Xitron (R(infinity)). TBW volumes were calculated using our modified BIS method (Jaffrin et al 2006 Med. Biol. Eng. Comput. 44 873-82) from R(infinity), R(t50) and from the fat-free mass measured by DXA, assuming a hydration rate of 73.2%. The same protocol and calculations were also carried out on a 2nd group of 21 subjects for independent validation. Data of the 1st group showed that values of R(t50), not significantly different from those of R(infinity), could be obtained by dividing R(50) by 1.231 in men and by 1.224 in women. Applying this method to the 2nd group yielded also values of R(t50) not significantly different from R(infinity). TBW volumes V(t50) obtained from R(t50) were not significantly different from those of our modified BIS method V(tn), or from TBW volumes obtained from DXA in both groups. A comparison with three BIA methods of TBW determination showed that our new method gave results in better agreement with TBW from DXA and from our modified BIS method.

A U-shaped bioartificial pancreas with rapid glucose-insulin kinetics. In vitro evaluation and kinetic modelling.

The lag in insulin release in response to glucose is an obstacle to the development of hybrid pancreatic devices, in which an artificial membrane protects transplanted islets against immune rejection. We designed a U-shaped bioartificial pancreas, in which the blood channel surrounds the islet chamber consisting of two flat membranes; blood circulates successively above the first membrane and then in the reverse direction, below the second membrane. Isolated rat islets were introduced into the chamber, which was perfused with Krebs buffer, and the kinetics of insulin release in response to glucose was determined. During a 20-min, 2.8-20-mM, square-wave glucose stimulation, insulin release in the effluent of the device rose from 0.7 +/- 0.2 to 3.2 +/- 1.0 ng/100 islets/min (P less than 0.05) within 3 min, and reached a maximal level of 12.8 +/- 3.3 ng/100 islets/min at 10 min; 5 min after the return of the glucose concentration to substimulatory level, insulin release dropped from 11.3 +/- 1.5 to 8.0 +/- 1.7 ng/100 islets/min (P less than 0.05), and reached basal value (1.0 +/- 0.2 ng/100 islets/min) 40 min after the end of the stimulation. A 0.1-mM/L/min ramp increase in glucose concentration triggered a significant rise in insulin release (P less than 0.02) when the glucose concentration reached 5.3 +/- 0.2 mM; islets concomitantly perifused within a chamber set up without membrane responded to the same glucose stimulation 5 min earlier. For up to 1000 islets, insulin release in response to glucose was linearly correlated to the number of islets (N = 12, P less than 0.01), indicating that insulin did not significantly inhibit its own secretion in this system. Finally, during glucose stimulation, the insulin concentration in the effluent from the chamber was found to be four times the concentration present at the turning point of the blood channel, suggesting that insulin was transferred into the perfusing medium in part by a countercurrent flux of ultrafiltrate crossing the membranes. We present herein the kinetic modelling of glucose and insulin transfer in this "ultrafiltration chamber," whose functional characteristics are compatible with closed-loop insulin delivery.

Aortic input impedance in normal man and arterial hypertension: its modification during changes in aortic pressure.

The purpose of this work was to study the factors determining aortic input impedance in hypertensive patients. Aortic input impedance (simultaneous measurements of aortic pressure and blood flow), mean (Wm) and pulsatile (Wp) powers and the Wp/Wm ratio were compared in normal subjects (n = 13) and hypertensive patients (n = 12) under basal conditions and during blood pressure manipulation--angiotensin infusion in five normal patients and nitroprusside infusion in six hypertensive patients. Pulse wave velocity (Möens-Korteweg equation; simultaneous measurement of aortic pressure and radius) was determined under basal conditions in normal subjects and in 11 hypertensive patients. The results show that: 1) the changes in impedance curves in hypertensive patients are related to increased peripheral resistance, pulse wave velocity, wave reflection and aortic radius; 2) in most hypertensive patients impedance curves are normalised when blood pressure is reduced, whereas the Wp/Wm ratio remains higher. This latter result demonstrates that pulsatile energy losses are greater in hypertensive patients and suggests either that the aortic wall remains stiffer, despite the reduction in aortic pressure, or that the flow wave becomes more pulsatile since impedance curves of hypertensive patients seen after lowering blood pressure are similar to those of normal subjects.

Pulsed flow cascade filtration. Long-term experience in low density lipoprotein and lipoprotein a removal.

The authors present the results of pulsed flow cascade filtration (PFCF) for low density lipoprotein (LDL) and lipoprolein a (La) removal in four patients with familial hypercholesterolemia. Forty-six treatments were performed. Pressure and flow pulsations were superimposed on the retentate of the secondary filter using a modified single roller peristaltic pump. The pulsation frequency was adjusted to achieve, on average, zero retentate flow (dead end filtration). The transmembrane pressure and sieving coefficients of immunoglobulin G, apolipoprotein A1, and apolipoprotein B of the secondary filter remained constant during filtration periods of 150 min. The PFCF technique was found to remove more LDL and La than did adsorption on dextran sulfate columns, while maintaining adequate albumin recovery (90%).

Extracellular and intracellular fluid volume monitoring during dialysis by multifrequency impedancemetry.

Impedance techniques can, in principle, permit non-invasive monitoring of extracellular (Ve) and intracellular (Vi) fluid volumes during dialysis. The authors present a method that determines the resistances Re and Ri of extracellular and intracellular compartments, respectively, by extrapolating impedance measurement toward zero, as well as infinite frequencies, according to the Cole-Cole model of biologic issues. These measurements were made using a XITRON 4000 B impedance meter (Xitron Technologies Inc., San Diego, CA) at frequencies ranging from 5 kHz to 1 MHz. The fact that the body is not a cylindric, homogeneous conductor is taken into account by introducing shape factors ke and ki and different resistivities pe and pi for the extracellular and intracellular fluid compartments. Assuming that these four unknown parameters can be regarded as constant during dialysis, the authors obtain: [formula: see text] where subscript o denotes initial values at start of dialysis. Impedances were measured at 30 min intervals on 11 pediatric patients and two adults, whereas total body water volume was determined by measuring urea in collected dialysate. Without ultrafiltration, Ve and Vi do not change significantly in percentage whereas, in the presence of ultrafiltration, Ve decreased by 15% to 25%. In cases when Vi does not change, it is possible to determine Ve and its variation during dialysis.

Comparison of optical, electrical, and centrifugation techniques for haematocrit monitoring of dialysed patients.

Haematocrits were measured as a function of ultrafiltration in a simulated haemodialysis circuit using bovine blood (plasma conductivity 12 mS cm-1) and hypotonic (8.6 mS cm-1) or hypertonic (16 mS cm-1) dialysates as well as in the absence of dialysate. A comparison was made between measurements by light absorption due to haemoglobin, by impedance in the blood line at 5 kHz using Hanai's model of blood conductivity, by conductivity measurements of blood samples at 1.2 kHz using a conductimeter, by centrifugation of blood samples and by calculations using fluid conservation. The validity of Hanai's model was verified to be satisfactory by direct blood and plasma conductivity measurements. In the absence of ionic transfer the impedance device underestimated the haematocrit by 5 to 7%. This underestimation reached 18% in the case of hypertonic dialysate, but this effect can be minimised if the haematocrit necessary for calibration is measured by centrifugation after 15 min of dialysate circulation when ionic balance is achieved. It was found that the optical method monitors haemoglobin concentration rather than red cell volume changes and is not affected by osmotic red cell swelling in the case of hypotonic dialysate. It can be concluded that the light absorption technique is both more accurate and more convenient to use than impedance.

Continuous measurements by impedance of haematocrit and plasma volume variations during dialysis.

A technique for continuous measurements of haematocrit and plasma volume in the arterial line of dialysed patients has been tested in vitro and in vivo. This method uses impedance measurements at 5 kHz and requires a single haematocrit measurement. It relies on two assumptions: that plasma resistivity does not change during dialysis and that blood resistivity obeys Hanai's model. Both assumptions are verified during in vitro tests. Haematocrits measured in vivo by this method are found to be in good agreement with direct measurements from blood samples. The haematocrit variation is then used to monitor changes in plasma volume, assuming conservation of erythrocyte volume. In addition, it is possible to obtain the variation in interstitial volume by combining these data with body impedance measurements.

Extra- and intracellular volume monitoring by impedance during haemodialysis using Cole-Cole extrapolation.

A method is presented for monitoring the relative variation of extracellular and intracellular fluid volumes using a multifrequency impedance meter and the Cole-Cole extrapolation technique. It is found that this extrapolation is necessary to obtain reliable data for the resistance of the intracellular fluid. The extracellular and intracellular resistances can be approached using frequencies of, respectively, 5 kHz and 1000 kHz, but the use of 100 kHz leads to unacceptable errors. In the conventional treatment the overall relative variation of intracellular resistance is found to be relatively small.

Plasma fractionation by dead-end membrane filtration: effects of membrane properties and plasma flux.

Plasma fractionation by membrane filtration permits the reinfusion of the patient with his own albumin. In this study, the influence of membrane nature and plasma flux on plasma fractionation in dead-end mode is investigated with acetate hollow fiber filters. It is found that transmembrane pressure TMP rises exponentially with time, the rate of increase being proportional to plasma flux. The faster TMP rises, the faster the drop in sieving coefficient SC. It is also found that albumin SC is a function of TMP and not of plasma flux. Theoretical analysis of the dead-end filtration was performed. This theoretical model indicates that the observed variation of TMP with time is consistent with the assumptions that pore volume decreases proportionally to the filtrate plasma volume.

Flows of red blood cell suspensions through narrow two-dimensional channels.

Apparent viscosity and mean channel hematocrits have been measured at various shear rates and feed hematocrits for red blood cell (RBC) suspensions flowing in two-dimensional channels. Three types of RBC were used in the suspensions : normal, partially hardened by heating at 50 degrees C and completely hardened by glutaraldehyde fixation. Channel height was varied from 20 to 200 mu and feed hematocrit from 5 to 55%. Measurements show that RBC deformability plays a dominant role in narrow channels and viscosity increases rapidly with decreasing cell deformability. Like in narrow tubes the apparent viscosity as well as the mean channel hematocrits decrease as the channel height is reduced. However the apparent viscosity in a channel remains slightly higher than the viscosity in a tube of diameter equal to the channel height. These results are consistent with the existence of a cell-depleted layer near the channel walls.

A process for enhancing selectivity and limiting plugging in plasmafractionation for APOB removal.

A new process using pressure and flow pulsations was investigated in the second stage of cascade filtration for ApoB removal. Due to enhanced filtration by the pulsations, the retentate flow vanished naturally without clamping the line, which reduced albumin and ApoA1 losses. The performance of Eval 4A and cellulose PF100 filters in this case were compared with that of classical dead end mode at a plasma flux of 18 ml/(min m2). The apparent sieving coefficient for small proteins (albumin, ApoA1) under pulsed conditions increased with increasing pressure instead of dropping as in dead end mode. In both cases the ApoB sieving coefficients remained below 0.02. These encouraging results were accompanied by a slower rise of transmembrane pressure showing a decrease of membrane plugging. Thus, the recovery of albumin and ApoA1 is increased while ApoB removal remains constant, which improves the selectivity of the fractionation stage.

In vitro study of combined convection- diffusion mass transfer in hemodialysers.

Clearance measurements have been obtained for urea, Vit B12 and myoglobin by simultaneous hemodialysis and ultrafiltration (UF) techniques, using various commercially available hemodialysers. The experimental results confirm that the overall clearance is less than the clearance of each process occurring separately due to interaction between convection and diffusion. Adequate urea clearance (100 ml/min) can be obtained either by using high ultrafiltration (60 ml/min) with low dialysate flow rate (70 ml/min) or moderate ultrafiltration (less than 30 ml/min) with higher dialysate flow rate (120 ml/min). The comparison of experimental clearance with the predictions of theoretical model (1) for urea are found to be in good agreement for dialysate flow rates of more than 300 ml/min and for smaller dialysate flow rates the predicted clearances are higher than measured ones. An approximate correlation for the overall clearance is CL = CLD + QF/2 where CLD is the dialytic clearance and QF the UF flow rate.

Reversibility of artifacts of fluid volume measurements by bioimpedance caused by position changes during dialysis.

The effect of temporary position changes, sitting up from supine, on extracellular (ECW) and intracellular (ICW) resistances and fluid volumes calculated from whole body bioimpedance using a Xitron 4200 impedancemeter was investigated on 8 patients during dialysis for a total of 11 tests. It was found that ECW resistance decreased instantaneously by an average of 2.3% when the patient sits up, due to plasma and interstitial fluid shift into the legs which decreases leg resistance, the major contributor to total resistance. This drop in resistance is incorrectly interpreted by the device as an increase in ECW volume which averages 235 ml. But this effect is completely reversible and both ECW resistance and fluid volume rapidly resume their normal course when the patient returns to his initial position. No significant variation in ICW resistance was observed in any of the patients at the position change. We conclude that segmental impedance, which has been proposed to minimize this artifact, is not advisable in dialysis monitoring and that it is simpler to ignore or switch off measurements during the position change so that later data are not affected by it.

Dynamic filtration of blood: a new concept for enhancing plasma filtration.

We have shown previously that blood flow pulsations created by intermittent squeezing of the inlet blood line significantly increased the plasma filtration rate in membrane plasmapheresis. However, in order to avoid hemolysis, the filtration increase had to be limited to about 50%. We have now devised a more efficient pulsation generator. By properly matching the tubing compliance and the pulsation amplitude, it is possible to extract 50 ml/min of plasma from 90 ml/min of blood at 36% hematocrit with a 1000 cm2 polypropylene hollow fiber filter without hemolysis. Simultaneous recording of the time course of plasma filtration rate measured by an electromagnetic flow meter and transmembrane showed that the increase in mean plasma flow rate was due to a dynamic filtration process which prevents the establishment of concentration polarization. The transmembrane pressure (Ptm) increases over a 0.5-second interval when the tube is squeezed. The membrane responds with an increase in filtration since the concentration polarization layer takes a few seconds to build up. The Ptm then drops when the tube is released before the polarization layer has time to build up appreciably and a sudden acceleration of the blood flow (velocity spike) helps clean the membrane, reducing the polarization. Tests with bovine show that the system is very efficient in reducing membrane plugging with small area filters.

Albumin recovery enhancement in membrane plasma fractionation using pulsatile flow.

In therapeutic plasmapheresis using cascade filtration, it is important to maximize albumin recovery while rejecting as many gamma-globulins as possible. Several membrane fractionation techniques were investigated using fresh bovine and human plasma and cellulose acetate filters (PF 100, AKZO). In dead end mode the sieving coefficients were found to decrease as transmembrane pressure increased. This was due to membrane plugging during the course of filtration after about 20 minutes which lead to a rapid increase in transmembrane pressure. In single pass mode the albumin recovery factor generally remains around 40% since the permeate flux is much less than the inlet flow. When strong pulsations (4 to 6 Hz) were superposed on the inlet plasma flow in single pass mode, the albumin sieving coefficient remained at about 0.95 while the permeate flux was increased by 106%. As a result a recovery factor of more than 80% could be sustained for at least 90 minutes without membrane plugging. Therefore pulsatile flow plasma fractionation seems to be an interesting approach to combine continuous operation with high albumin recovery.

Hemolysis reduction in plasmapheresis by module design: operating with pulsed flow filtration enhancement.

This study is an investigation into the effects of module design on hemolysis levels during the filtration enhancement of plasmapheresis using pulsed blood flow. Two polypropylene hollow fiber modules (F1 and F2) were compared under steady and pulsed flow conditions. The fiber lengths and membrane areas of the respective filters were 136 mm, 0.1 m2 and 226 mm, 0.25 m2 and were used at wall shear rates of 400 and 600 s-1. Fresh citrated bovine blood was circulated at 90 ml/min through their fibers with permate and retentate being recirculated. Plasma samples were analysed by the Cripps method to obtain the mean free hemoglobin concentration. A comparison of the average hemoglobin released with time per fiber, and per unit length allowed an assessment of the effects of fiber length and shear rate on hemolysis levels. It was concluded that a reduction in area and fiber length would reduce hemolysis levels and design equations are suggested to find the optimum length.

Comparison of polyamide and polypropylene membranes for plasma separation.

Plasma separation experiments were made with polyamide experimental prototype hollow-fiber plasma filters with surface areas between 0.025 m2 and 0.1090 m2 using bovine blood collected in acid citrate dextrose (ACD). The maximum filtration velocity rose with the wall shear rate gamma w as gamma w 0.72 +/- 0.02 and decreased with the length of fiber L as L-0.41 with a correlation coefficient of 0.97 +/- 0.02. The results were similar to those with polypropylene fibers. We also investigated the occurrence of hemolysis as a function of shear rate and transmembrane pressure. The free hemoglobin concentration of filtered plasma was checked using a U.V. spectrophotometer. It was concluded that polyamide membrane filters can be safely used for plasma separation from blood.

Urea, creatinine and phosphate kinetic modeling during dialysis: application to pediatric hemodialysis.

The kinetics of urea, creatinine and phosphate removal during dialysis were investigated in pediatric patients using a two-pool model taking into account fluid shifts and mass transfer between the two compartments. It is found that even urea must be described by a two-pool model since it presents a post dialysis rebound due to equilibration between the two compartments. Phosphate plasma concentration drops very sharply during the first hour of dialysis and rises rapidly during the rebound period. This pattern cannot be accounted for by the classical two-pool model with constant generation rate and mass transfer coefficients, but corresponds to a large time-dependent phosphate influx from the intracellular compartment in which phosphate is generated by biochemical reactions or liberated from the bones. This influx was calculated for four patients representing 8 dialysis sessions and was found to reach a plateau after 90 minutes of dialysis, dropping rapidly during the rebound period.

Mass transfers in a fluidized bed bioreactor using alginate beads for a future bioartificial liver.

Fluidized bed bioreactor with alginate beads may be an alternative to hollow fiber cartridge to host hepatocytes for bioartificial liver purposes. After the bioreactor design and the characterization of fluid mechanics, the present study was aimed at analyzing bi-directional mass transfers of calibrated species between external fluid and empty beads. Static (batch) and dynamic (fluidized bed bioreactor) experimental conditions were analyzed. A simple modelling approach permitted the definition of mass transfer coefficients. The motion of beads within the bioreactor clearly enhanced mass transfer kinetics, but did not alter the amount exchanged. The shear enhanced diffusion coefficient for VitB12 was 20 times higher in the fluidized bed bioreactor than under batch conditions, proving the efficiency of such a device.

Modelling of plasma-separation through microporous membranes.

Available mathematical models of ultrafiltration have been used to predict changes in maximum plasma filtration rate with wall shear rate for given filters and blood properties. We have done many plasmapheresis experiments in vitro, using hollow-fiber filters (500-1000 cm2) and fresh bovine blood collected on ACD or heparin. The comparison between predicted and experimentally obtained filtration rates was good for models based on the concentration polarization theory and lift velocity theory. In other experiments with pulsatile inlet flow we found that plasma filtration rate increased by 20 to 50% compared to non-pulsatile conditions. These results are in good agreement with the modified model of ultrafiltration incorporating pulsating flow. This paper presents relationships between plasma filtration velocity (steady and pulsating flow) and hemolysis limit as a function of wall shear rate and filter size.