Sieving and Reflection Coefficients for Sodium Salts and Glucose During Peritoneal Dialysis in Rats1

Transcription

1 Sieving and Reflection Coefficients for Sodium Salts and Glucose During Peritoneal Dialysis in Rats1 Tzen Wen Chen, Ramesh Khanna,2 Harold Moore, Zbylut J. Twardowski, and Karl D. Nolph T,w. Chen, R. Khanna, H. Moore, Z.J. Twardowski. K.D. Nolph. Division of Nephrology. Department of Medicine. and Dalton Research Center. University of Missouri-Columbia, Harry S. Truman Veterans Administration Hospital-Columbia. Columbia, MO (J. Am. Soc. Nephrol. 1991; 2: ) ABSTRACT The two-part studies reported herein address peritoneal membrane ultrafiltrate (UF) characteristics during peritoneal dialysis exchanges in rats. In the studies of part I, the sieving coefficients for sodium, chloride, and total solutes during hydrostatic UF after instillation of rat serum into the peritoneal cavity of rats were calculated. Thirty-six rats were divided into six groups (N = 6) according to the following peritoneal dialysis exchange cycle times: 60, 120, 180, 240, 480, and 960 mm. Thirty milliliters of pooled rat serum were infused i.p. with the animal being conscious except during infusion and drainage. The study showed in the early phase of exchanges, when oncotic and osmotic pressure gradients were absent, net UF presumably due to capillary hydrostatic pressure and sodium sieving during such UF. Sieving coefficients for sodium (0.72), chloride (0.77) and total solutes (0.73) were determined by using standard formulae. In the second part of these studies, the kinetics of fluid movement after the instillation of 5% dextrose solution into the peritoneal cavity of rats were analyzed. A very low UF rate was observed early in the exchange when the glucose gradient between the dialysis solution and blood was at its peak. The UF rate gradually increased as the sodium entered the dialysis solution from the blood. At the time of low UF rate with high glucose gradient, presumably the osmotic pressure generated by the glucose in the dialysis solution was countered by the osmotic pressure of solutes in plasma, i.e., sodium and its I Received January Accepted July 22, Correspondence to Dr. R. Khanna, MA 436 School of Medicine, University of Missouri-Columbia, Columbia, MO I / $03.00/0 Journal of the American Society of Nephroiogy Copyright C 1991 by the American Society of Nephrology anions. Under such conditions, the ratio of concentration gradients between the blood and dialysis solution of sodium salt and glucose reflects the ratio of reflection coefficients of corresponding solutes. The calculated ratio of reflection coefficients for sodium salts and glucose for the peritoneal membrane in our study was The reflection coefficient for glucose was derived to be 0.32, approximately similar to previously reported values. These studies showed that solute sieving could occur during what was believed to be hydrostatic UF during peritoneal dialysis exchanges and thus is not a phenomenon unique to osmotic pressure UF. During these simple studies, approximate sieving and reflection coefficients for various solutes that are present in the blood and dialysis solution were calculated. The calcu- Iated values were similar to previously reported values derived during complex mathematical models. Key Words: Hydrostatic ultrafiltration. osmotic ultrafiltration, transcapillary ultrafiltration, peritonealcavitylymph flow, peritoneal membrane T he ability of a water-soluble solute to exert an effective osmotic pressure gradient across a membrane is a function of its molecular radius rebative to the radius of the water-filled pores in that membrane. If the membrane is perfectly semipermeable (permeable to water, but not to the solutes), the effective and the theoretical osmotic pressures are equal. However, for membranes that are not perfectly semipermeable, such as the penitoneab membrane, the effective osmotic pressure will be less than the theoretical value. The effective osmotic pressure divided by the theoretical osmotic pressure has been defined by Staverman (1) as the reflection coefficient o At present, there is limited information about reflection coefficients for the peritoneal membrane and various solutes that are present in penitoneal dialysis solutions. Direct measurement of the penitoneab membrane reflection coefficient for a given solute is not possible at present. Even indirect measurements are not easy and are subject to criticism. What is usually measured clinically, however, are estimates of peritoneal membrane solute sieving coefficients. The sieving coefficient for any solute is the ratio of the concentration in the ubtrafiltrate to that in the plasma water. It can have a maximum value of 1.0 (no membrane sieving) and a minimum value of Volume 2. Number

2 Chen et al (total sieving). Sieving coefficient values for a variety of sobutes for the penitoneal membrane have been estimated and previously published (2). These are usually underestimates of true sieving coefficients, because sieving creates gradients for solute net diffusive transport above that created by simple convection; thus, they should be considered as a net sieving effect. The true sieving coefficient (5) for a solute when subtracted from 1.0 yields a value that approximates the reflection coefficient (tr): 7 : = i - S. This value is called or the rejection coefficient. Sodium sieving is observed during hypertonic peritoneal dialysis exchanges where water can be removed from the extraceblular space without proportional amounts of extracellular sodium; this can cause hypernatremia (3). It is as yet unclear whether solute sieving is a membrane function or a feature peculiar to certain types of ultrafiltration (UF), especiably with a permeant solute such as glucose used as the osmotic agent. In vitro studies in hollow fiber dialyzers have examined solute sieving coefficients during UF with hydrostatic pressure, osmotic pressure with a nonpermeant solute, and osmotic pressure with a permeant solute (4). The batter was associated with lower sieving coefficients than the first two, and the first two gave similar results. Under such conditions, it was proposed that molecular interaction within the membrane impaired convective transport, causing solute sieving. Alternatively, it has also been proposed that hydrostatic or osmoticinduced water flow may occur transcellulanly without sodium and that the sodium that is removed accompanics the portion of ultrafiltrate that is generated through intercellular channels (5). Therefore, sieving coefficients should be similar regardless of whether the pressure is hydrostatic or osmotic. However, the inability to study the kinetics of hydrostatic-induced UF during peritoneab dialysis has made it difficult to compare hydrostatic and osmotic solute sieving coefficients across the peritoneum in viva. In the studies reported herein, we have analyzed the kinetics of UF after the instillation of pooled rat serum into the penitoneal cavity of rats. Osmotic and oncotic pressure gradients are thus negligible at the beginning of such exchanges. Any net UF occurring during the early phase of such exchanges is presumed to be hydrostatically induced. In this model, we were interested in seeing whether sodium sieving could be observed during UF across peritoneab capillaries in the absence of osmotic pressure. If hydrostatic-induced UF was associated with sodium sieving under the conditions of our study, we predicted that the dialysate sodium concentration would fall because of the entry of a low-sodium ultrafiltrate into the penitoneab cavity. Such an observation would essentially counter the concept that solute sieving is a phenomenon peculiar to osmotic-induced UF and would support the hypothesis that water flow across the penitoneal membrane from capillaries to the peritoneal cavity traverses some pathways where sodium cannot foblow. The mobility of solutes through a semipermeable membrane is affected by the Gibbs-Donnan effect (6). Therefore, we calculated the net solute sieving coefficients on the basis of the solute concentrations in the ultrafiltrate and blood, correcting for Gibbs- Donnan effect on electrolyte distribution. When cabculating the net solute sieving coefficients, many previous studies have neglected contributions of the Gibbs-Donnan phenomenon. In the studies reported herein, by infusing pooled rat serum into the peritoneal cavity of rats, we maintained equal concentrations of protein on both sides of the penitoneal membrane. Under such conditions, the Gibbs-Donnan effect of plasma proteins on serum electrolytes was minimal because of similar protein concentrations in the infused pooled serum. In the second part of the studies, we have further analyzed the kinetics of fluid movement after the instillation of 5% dextrose solution with and without electrolytes into the penitoneal cavity of rats. In the absence of electrolytes in the 5% dextrose solution, UF in the early moments of the exchange should reflect mainly the net balance of osmotic forces gencrated by glucose and sodium and its anions. Presumably, the osmotic pressure generated by the glucose in the diabysate would be countered by the osmotic pressure generated by the plasma sodium and its anions, the predominant osmotically active salutes in the plasma. We predicted the UF rate would be initially lower in a 5% dextrose exchange until rapid sodium salt entry into the peritoneal cavity increased dialysate sodium concentration and, thereafter, UF would increase as glucose forces were less opposed. Because the UF rate is a function of ( 1 ) the net sum of the products of osmotic gradients and respective solute reflection coefficients for all water soluble solutes and (2) membrane characteristics and because in our model the membrane characteristics would be similar for the two sets of experiments, the ratio of reflection coefficients for sodium salts and glucose for the peritoneal membrane could be calcubated from the ultrafiltration kinetics in these studies. Materials and Methods Part I Thirty-six male Sprague-Dawbey rats were divided into six groups (N = 6), corresponding to the following cycle times: 60, 120, 180, 240, 480, and 960 mm. Journal of the American Society of Nephrology I 093

3 Sieving and Reflection Coefficients for Sodium and Glucose The animals were anesthetized with methoxyflurane (MetofaneR; Pitman-Moore, Inc., Washington Crossing, NJ) in a large jar and were placed supine on a 37#{176}Cheating pad. Thirty milliliters of pooled rat serum (Stellar Bio Systems, Inc., Rockville, MD) was injected into the penitoneal cavity with a 2 1 -gauge needle within 2 mm. A blood sample (2 ml) was taken through direct cardiac puncture. After these procedures, the animals were allowed to wake up, were placed back in their cages, and were allowed free access to food and water. The animals were reanesthetized with methoxyfluranejust before the end of the cycle and were placed supine on a 37#{176}Cheating pad. An indwelling peritoneal catheter was inserted through a midline incision 1 cm below the xiphoid process. The dialysate was drained as completely as possible. The abdomen was then opened, and all residual fluid (residual volume) was removed by aspiration. A poststudy blood sample (2 ml) was taken through direct cardiac puncture. Part II Section A. Twenty-one male Sprague-Dawley rats were divided into seven groups (N = 3) corresponding to the following cycle times: 15, 45, 120, 240, 480, 960. and 1,200 mm. The animals were anesthetized with methoxyflurane (MetofaneR; Pitman-Moore, Inc.) in a largejar and were placed supine on a 37#{176}C heating pad, and 30 ml of 5% dextrose solution containing 2.5 g of albumin was injected into the penitoneal cavity with a 2 1 -gauge needle within 2 mm. The rest of the protocol was the same as that described for part I. Section B. After the above experiments were completed, in a separate study, another 2 1 rats were injected i.p. with 5% dextrose DianealR (Baxter Healthcare Corp., Deerfield, IL) solution (commercial 4.25% Dianeal dialysis solution made 5% by the addition of dextrose) containing 2.5% albumin as per the part II, section A study. The rats were divided into seven groups corresponding to cycle times simibar to those in the section A studies. The rest of the protocol was the same as that of section A. The aim of this part of the study was to confirm whether the transcapillary UF (TCUF) rate during the initial phase of the exchange with electrolyte-containing dialysis solution was indeed identical to that extrapolated to time zero from the late exponential curve of the 5% dextrose-electrolyte free exchange. Biochemical Assays All serum and dialysate samples were assayed as follows. Osmolality was determined with a Wescor 5100 B vapor point osmometer (Wescor Inc., Logan. UT). Albumin concentrations in serum and dialysate were measured by a modified bromcresol green method (7). Protein concentrations were measured by the biuret method. Electrolytes and glucose concentrations were determined on the ASTRA-8 system (Beckman Instruments, Inc., Brea, CA). Diabysateglucose concentrations were determined by the O-tobuidine method. Calculations Corrected Solute Concentrations. All electrolyte concentrations in serum were corrected to their approximate concentrations in the water portion of the solution by the relationship: Corrected concentration = (observed concentration) X x Pr where Pr equals the protein concentration in each sample in grams per 1 00 milliliters. UF, Albumin Clearance, and TCUR rates. Peritoneal cavity lymphatics drain i.p. fluid by bulk transport without an increase or decrease in protein content (8,9), and i.p. macromolecules of molecular weight >20,000 are returned to the venous circulation almost exclusively by peritoneal lymphatics (1 0, 1 1 ). Some feel that convective movement of fluid and albumin directly across the peritoneal membrane may also contribute substantially to albumin clearance from the peritoneal cavity (1 2). In reality, both sites may be involved in the convective transport of fluid and protein from the peritoneal cavity. The exact proportion contributed by each is as yet not clearly established. Regardless of the exact site of the convective transport of fluid and protein from the penitoneal cavity, the albumin clearance (AC) for the purpose of this study represents the albumin cleared from the peritoneal cavity. The site of clearance is of no consequence to the interpretation of our results. Thus, AC during penitoneab dialysis can be calculated from the rate of disappearance of albumin from the peritoneal cavity. Because the net amount of albumin that could move, if at all, from the capillary blood into the penitoneal cavity in the absence of an oncotic pressure gradient is unknown, as in the study presented here, and because the amount that has been documented to move into the penitoneal cavity during a protein-free fluid exchange is very small compared with what was present in the i.p. infused solution in our study animals, we elected to neglect any net protein entry into the penitoneal cayity during study exchanges when calculating the abbumin clearance from the peritoneal cavity. TCUF is defined as the net volume of bidirectional transcapillary water movement into the peritoneal cavity during a time interval. A negative value for TCUF represents essentially transcapilbary fluid absorption (backfibtration). TCUF is equivalent to meas Volume 2 Number

4 Chen et al ured UF plus the amount of convective loss of fluid from the peritoneal cavity over a time interval. TCUF can also be estimated from the dilution of the initial dialysate albumin concentration, assuming that the i.p. albumin concentration is essentially unchanged by the convective AC of i.p. fluid and that any decrease in the albumin concentration results from the net flux of fluid from the peritoneal microcirculation. Net UF or absorption represents the increase or decrease in i.p. volume over a time interval, respectively. Thus, (1 ) TCUF (milliliters) at time t (hours) = Co/C(IPV0) - (IPV0); (2) AC during a given dwell time t (hours) = C/C(IPV0) - C/Cg(IPVj; and (3) measured net UF (milliliters) at time t (hours) = (IPV) - (IPV0), where Co and C are dialysate albumin concentration at time 0 and time t (hours), respectively; Cg is the geometric mean of dialysate albumin concentration at time 0 and time t (the geometric mean [CJ closely represents the time-averaged i.p. albumin concentration and allows calculation of AC with two data points); IPV0 is the i.p. volume at time 0, or infusion volume (we have previously found preequilibration residual volume in our rat model to be 1 ml or less and this was neglected in calculations); and IPV is the i.p. volume at time t (hours), or drain volume plus sample volume. AC from the peritoneal cavity is assumed to represent an estimate of convective fluid and protein absorption rates from the peritoneal cayity and to reflect, at least in part, lymphatic absorption. Sodium, Chloride, and Osmolar Concentrations in the Ultrafiltrate. Because the sodium, chloride, and osmolar concentrations in the capillary blood and i.p. infused serum are similar in the current study model, no significant net diffusion is assumed to occur during the exchange. Under such a condition, the sum of sodium in the drained fluid (after a dwell time of t) and the sodium that left the peritoneal cavity with convective backabsorption of fluid during the time t should equal the amount of infused sodium. Any additional sodium in the dialysate at time t would be a measure of the sodium that entered the peritoneab cavity with the ubtrafiltrate fluid. Thus: Na = Na + Na - NaAC and rearranging the above equation would yield: Na,,, = Na + NaAc - Nao where Na is the amount of sodium in the drained fluid at time t, Na is the sodium in the infused solution, NaAc is the convective loss of sodium during time t, and Na represents the amount of sodium in the ultrafiltrate during time t. The concentration of sodium in the ultrafiltrate is calculated by dividing the amount of sodium by the volume of transcapillary ultrafiltrate during the time t: or UNa - - Una - Na + NaAC TCUF - Na0 (Na, x Vt) + (Nag X AC) - (Na0 X Vo) TCUF where UNa is the sodium concentration (milliequivalents per liter) in the ultrafiltrate, Na is the dialysate sodium concentration (milliequivalents per liter) at time t, V, is the drainage volume (milliliters) at time t, Nag is the geometric mean of sodium concentration (milliequivalents per liter) during the exchange, AC is the AC rate (milliliters per minute) from the pentoneal cavity during time t, TCUF is the volume of total ultrafiltrate (liters) during time t (sum of drainage volume and AC during time t), Na is the dialysate sodium concentration at time 0, and V0 is the volume of infusion. Sieving Coefficient of Solutes. Ratios of solute concentrations in the ultrafiltrate and plasma are used to calculate sieving coefficients (S) for solutes during the periods of UF and backfiltration. For sodium, the ratio would be 0 UNa Na rna where U8 is the sodium concentration in the ultrafiltrate and PNa is the sodium concentration in the plasma. Ratio of Reflection Coefficients of Sodium Salt and Glucose (UN&/U) for the Peritoneal Membrane. Because TCUF is determined by osmotic, oncotic, and hydrostatic pressure gradients, solute reflection coefficients, and membrane characteristics, the equation for TCUF across the penitoneal membrane during a peritoneal dialysis exchange may be written as follows: TCUF = osm). M + (Onc)..f- HP. K where osm is the crystalloid osmotic gradient, osm is the mean reflection coefficient of crystalloid sobutes, Onc is the protein oncotic pressure gradient, Pr is the protein reflection coefficient, LHP is the hydrostatic pressure gradient across the membrane, and K is the membrane area and permeability constant. At 0 TCUF, these factors on two sides (blood and peritoneal cavity) of the peritoneal membrane balance each other evenly: 1(osM). l1em + (Onc). Pr + LHP. K = 1(osM). ToaM + (Onc). Pr + LHPJ. K Sodium salts are the major ions contributing to the generation of osmotic force from the blood (contri- Journal of the American Society of Nephrology I095

5 Sieving and Reflection Coefficients for Sodium and Glucose butions to osmolality by solutes such as calcium, potassium, magnesium salts, and glucose are minimal). When dextrose solution is infused into the peritoneal cavity and at 0 TCUF: and salt DNa sait). UNa salt + (Onc). Pr + HP. K then: = (Dg Pg). ag + (zonc). apr + HP. K where salt is the concentration of sodium salt (milliequivabents per liter) in the serum at the time when TCUF is 0, DNaI( is the concentration of sodium salt (milliequivalents per biter) in the infused solution at the time when TCUF is 0, ana salt is the reflection coefficient for sodium salt, Dg is the glucose concentration (micromoles per biter) in the infused solution at the time when TCUF is 0, and Pg is the glucose concentration (micromobes per liter) in the serum at the time when TCUF is 0; because the hydrostatic and oncotic pressure gradients and membrane constants are similar on both sides, the equation can be simplified as follows: (PNa salt - DNa sa1t) Na salt (Dg - Pg). Nasalt Dg - Pg Flow rates (mi/mm) Ultrafiltration 4 1 Peak TCUF rate Backfiltration Mean AC rate e- TCUF rate Mean AC rate Figure 1. TCUF and AC rates during 960-mm peritoneal exchanges with rat serum. Concentration (meq/l) Osmolality (mosm/kg) -c ± -i I g PNa salt - DNa salt Statistics All values are expressed as mean ± SE, with selected means compared by the signed rank test or the Mann-Whitney test wherever appropriate. RESULTS Part I After instillation of near isosmotic-isoncotic rat serum into the penitoneal cavity, TCUF and AC rates at the previously given time intervals are depicted in Figure 1. During the first hour, the TCUF rate was low ( ml/min) and then steadily increased, reaching a peak rate of ml/min at 1 20 mm; thereafter, the rate decreased to near zero by 720 mm. The rate was negative thereafter with an average rate of ml/min. The mean AC rate (cxcluding the rate during the first 60 mm allowing for albumin absorption to surrounding tissues) was 0.02 ml/min. Cumulative AC volume was 19.2 ml over 960 mm. The movement of fluid between the capillanes and the penitoneal cavity can be said to occur in two phases: the initial UF phase (presumably due to hydrostatic forces, because osmotic and oncotic pressure gradients were negligible) and the later backabsorption phase (due to the presence of counteracting sodium-osmotic and protein oncotic pressure gradients) Na-Serum -8- Ci-Dialyaste Na-Diaiysate * Cl-Serum - Osm-Serum -4-- Osm-Dialysate Figure 2. The mean dialysate and serum sodium, chloride (milliequivalents per liter) and osmolality (milliosmolal per kilogram) during 960-mm exchanges with rat serum. The mean dialysate and serum sodium and chloride concentrations and osmolality are shown in Figure 2. Dialysate sodium concentrations steadily decreased to a low of 144 ± 0.8 meq/l at 240 mm from 148 ± 0.7 meq/l at infusion. The changes in the concentrations of diabysate chloride and osmolality essentially paralleled the sodium changes. The calculated concentrations of sodium and chboride and osmolality in the ultrafiltrate during the early phase and in the backfiltrate during the later phase at the previously given time intervals were not significantly different, and their mean values are given in Table 1. The calculated sieving coefficients for sodium, chloride, and total salutes at the previousby given time intervals were also similar, and Table 2 shows their corresponding mean values. Part II Section A. With 5% dextrose-only exchanges, TCUF rates at the previously given time intervals are 285 I096 Volume 2 Number

6 Chenetal TABLE 1. Concentrations of sodium, chloride, and total solutes in the ultrafiltrate and backfiltrate during peritoneal dialysis exchanges in rats Na C1 Na+Cl Total Solutes Osmolality Gap (meq/l) (mosm/kg) (mosm/kg) Ultrafiltrate 113±6 87± ±36 18 Backfiltrate TABLE 2. Mean net sieving coefficients of sodium, chloride, and total solutes during peritoneal dialysis exchanges in rats Na Cl Total Solutes Ultrafiltra ± ± ± 0.12 tion Backfiltra tion n TCUF rate (mi/mm) D B-- 5% Dextrose Figure 3. TCUF rates during 1,200 mm of peritoneal exchanges with 5% dextrose. shown in Figure 3. During the first 1 5 mm, when the ratio of the dialysate to plasma glucose is the highest (Figure 4), the TCUF rate was low at ml/min. Then, the rate steadily increased, reaching a peak rate of ml/min at 1 20 mm; thereafter, the rate decreased, coming to a halt at about 720 mm. The rate was negative thereafter with an average rate of ml/min. Dialysate-to-plasma ratios (D/P) of sodium and osmolality are shown in Figures 5 and 6. The peak TCUF rate of at 1 20 mm was reached near peak D/P osmolality. The sodium D/P ratio at 120 mm was The early increase in D/P osmolabity was mainly due to an increase in dialysate osmolality with rapid influx of sodium by diffusion. Predicted TCUF rates (by extrapolation of the curve in Figure 5 to 0 time) if a 5% dextrose electrolyte solution were to be used instead of a 5% dextrose solution are shown in Figure 7. The predicted TCUF rate at 0 time would be 0.24 ml/min. The calculated ratio of the sodium salt to glucose reflection coefficients was 0.85 at time 0 when TCUF was near 0. In part I of this study, the sieving coefficient for total sobutes was calculated to be 0.73 for the penitoneal membrane. By substitution, the totai osm = 1 - S would yield a reflection coefficient value of 0.27 for the total solutes in the blood. Because sodium salts are the predominant ions in the blood, the reflection coefficient values for them were assumed to be almost similar to total solutes: osm 1 - SosM = = 0.27 Nasalt Nasalt = G = Section B. In a separate study, the TCUF rates at the previously given time intervals were measured with a 5% dextrose Dianeal#{174} solution and the TCUF rate curve was superimposed on the curve obtained with a 5% dextrose solution (Figure 8). The actual curve is almost identical to the predicted curve drawn on the basis of extrapolation, supporting our hypothesis that lack of sodium salt in the 5% dextrose solution was probably the cause of low UF during the early phases of exchange; sodium salts exert osmotic pressure across the peritoneal membrane and essentially neutralize the osmotic pressure of dextrose early in the 5% dextrose in water exchange. Only when sodium salt gradients across the membrane begin to dissipate does TCUF begin. Figure 9 shows that changes in D/P glucose over time were similar with or without electrolytes in the instilled solution. Figures 1 0 and 1 1 allow direct comparison of D/P osmolality and sodium during the different types of exchanges, respectively. DISCUSSION In part I of our study, we observed a low electrolyte ultrafiltrate in the early phases of the exchange with a decrease in dialysate sodium concentration and Journal of the American Society of Nephrology 1097

7 Sieving and Reflection Coefficients for Sodium and Glucose DIP Glucose % Dextrose Figure 4. Glucose gradient between the dialysate and plasma during the 5% dextrose study. D/P Na I o % Dextrose 5% Dextrose #{149}NaCI Figure 5. Sodium gradient between the dialysate and plasma during the 5% dextrose study. -a-- TCUF rate TCUF (extrapolatmon) Figure 7. Predicted TCUF rate by extrapolation to time zero from the exponential TCUF curve (from Figure 3) of the 5% dextrose-electrolyte-free solution exchange. 1.2 DIP Osmolality 0.24 TCUF rate (mi/mm) o % Dextrose 5% Dextrose 5% Dextrose #{149}NaCi Figure 6. Osmolar gradient between the dialysate and plasma during the 5% dextrose study. Figure 8. Comparison of TCUF rates during exchanges with 5% dextrose with and without electrolytes Volume 2 Number

8 Chen et al D/P 15 Glucose D/PNa % Dextrose 5% Dextrose #{149}NaCi L _- -_ % Dextrose Figure 9. Comparison of dialysate-to-plasma glucose ratios during exchanges with 5% dextrose with and without electrolytes. Figure 1 1. Comparison of dialysate-to-plasma sodium ratios during exchanges with 5% dextrose with and without electrolytes D/P Osmohty 0.8 L % Dextrose. -5% Dextrose #{149}NaCi Figure 10. Comparison of dialysate-to-plasma osmolality ratios during exchanges with 5% dextrose with and without electrolytes. osmobality. This ultrafiltrate was presumed to be hydrostatically induced because the oncotic and osmotic pressure differences between the fluids in the penitoneal cavity and capillaries were negligible. The fact that this hydrostatic-induced UF was low in sodium suggests that the webb-documented sodium sieving during hypertonic penitoneal dialysis can occur with hydrostatic UF. Molecular interaction within the membrane has been proposed to explain the increase in sieving with osmotic-induced UF with a permeant solute (4). Our observation of sodium sieving with what is most likely hydrostatic UF suggests that sieving of electrolytes is a feature of membrane and not osmotic-induced UF per Se. Water flow induced by hydrostatic or osmotic pressure may occur transcelbubarby without sodium salts or through very small intercellular channels with surface charges where electrolyte movement is impeded (5). We observed similar mean sieving coefficients for sodium, chloride, and total sobutes during the early UF and later backfiltration phases of dwell. The ---- mean sieving coefficient of 0.72 ± 0.04 for sodium in this study, as predicted, is higher than previously reported values of 0.54 ± (1 3) and 0.56 ± 0.04 (1 4). The former (1 ) was a study done during a 30- mm dwell, (2) assumed lymph flow was negligible, (3) did not correct the sodium concentration for plasma water, and (4) more importantly, studied calculated sieving during glucose-induced UF. The lower value observed in the study of Rubin et a!. (1 4) may have been because of the higher osmotic gradient employed to induce UF, because decreases in the net sieving coefficient could occur with higher UF rates (1 5, 1 6). In addition, net sieving during UF may depend upon multiple factors such as dialysate flow rate, concentration gradients between two sides of a membrane, temperature, surface area, dialysate volume, and membrane differences from species to species. Any of the variables mentioned herein could have been a factor in observed differences between studies. Besides, none of the studies cited above took Gibbs-Donnan effects into consideration when calculating sieving coefficients. Gibbs-Donnan effects reduce the sieving coefficient of cations and increases that of anions. In part II of the study, we observed a very low TCUF rate initially with the 5% dextrose solution although the glucose osmotic gradient was at its maximum. As dialysate osmolality (predominantly due to the influx of sodium salts) increased to near serum levels, the TCUF rate increased. Osmotic force generated by the dialysate glucose must have been counterbalanced by the sodium salt gradients early in the exchange (when in fact the sodium salt gradient was at its maximum), acting in the opposite direction. Gradual diffusion of sodium into the dialysate, reducing the counterbalancing sodium osmotic force, can explain the increased TCUF during the second hour. The fact that the TCUF rates at and after 1 20 mm were similar Journal of the American Society of Nephrology I 099

9 Sieving and Reflection Coefficients for Sodium and Glucose for 5% dextrose and 5% dextrose Dianeab#{174} solutions supports our hypothesis that the difference in TCUF rates between the two solutions was initially due to the osmotic pressure exerted by the opposing salt and glucose gradients. On the basis of the assumption of near neutralization of forces at time 0, we calculated the ratio of sodium to glucose reflection coefficients to be Further, we calculated the 1 - S and sieving coefficient for glucose as 0.32 and 0.68, based on our sodium sieving coefficient of 0.72 in part I of the study. A previous study in rats by a series of complex computer simulations with a pharmacokinetics model determined the glucose reflection coefficient for the peritoneab membrane during isosmotic and hypertonic exchange to be 0.37, a value not too different from the value of 0.32 determined in our study (17). In summary, we have shown in our studies sodium sieving during probable hydrostatic UF suggesting the phenomenon to be a membrane characteristic rather than a feature of osmotic-induced UF. We calculated the sieving coefficients of sodium and chloride for the rat penitoneab membrane to be 0.72 and respectively, and the ratio of reflection coefficients of sodium salt and glucose for the peritoneal membrane to be ACKNOWLEDGMENTS This study was supported by the Division of Nephrology research fund. The authors acknowledge the secretarial assistance provided by Jeffrey Smart. REFERENCES 1. Staverman AJ: The theory of measurement of osmotic pressure. Recueil 1951:70: Henderson L. Ultrafiltration with peritoneal dialysis. In: Nolph KD, ed. Peritoneal Dialysis. 2nd Ed. Boston: Martinus Nijhoff Publishers; 1985: Ahearn DJ, Noiph KD: Controlled sodium removal with peritoneal dialysis. Trans Am Soc Artif Intern Organs 1972;18: Twardowski ZJ, Nolph KD, Popovich R, Hopkins CA: Comparison of polymer, glucose, and hydrostatic pressure induced ultrafiltration in a hollow fiber dialyzer: Effects on convective solute transport. J Lab Clin Med 1978;92: Nolph KD, Hopkins CA, New D, Antwiler GD, Popovich RP: Differences in solute sieving with osmotic vs. hydrostatic ultrafiltration. Trans Am Soc Artif Intern Organs 1976;22: Schneck DJ (ed). Engineering Principles of Physiologic Function. New York: New York University Press; 1990: DMA, Inc. Albumin procedure. Methods Brochure. Arlington, TX: DMA, Inc.: Henriksen Jil, Lassen NA, Parving H, Winkler K: Filtration as the main transport mechanism of protein exchange between plasma and the penitoneal cavity in hepatic cirrhosis. Scand J Clin Lab Invest 1980;40: Nicoll PA, Taylor AE: Lymph formation and flow. Annu Rev Physiol 1977;39: Arnie 5: Transpenitoneal exchange. IV. The effect of transperitoneal fluid transport on the transfer of solutes. Scand J Gastroenterol 1970;5: LIII SR, Parsons RH, Buhac I: Permeability of the diaphragm and fluid absorption from the penitoneal cavity in the rat. Gastnoenterobogy 1979;76: Flessner ML Dedrick R, Rippe B: Letter to the editor. ASAJO Trans 1 989;35: Nolph KD, Hano JE, Teschan PE: Peritoneal sodium transport during hypertonic peritoneal dialysis. Ann Intern Med 1969;70: Rubm J, Klein E, Bower JD: Investigation of the net sieving coefficient of the peritoneal membrane during penitoneal dialysis. ASAIO J 1982;5: Nolph KD, Ahearn DJ, Esterly JA, Maher JF: Irreversible, morphological and functional changes in hollow fiber kidneys with a single dialysis. Trans Am Soc Artif Intern Organs 1 974;20B: Ambard L, Trautman S. Ultrafiltration. Springfield, IL: Charles C. Thomas; 1960: Daniels FH, Leonard EF, Cortell S: Glucose and glycerol compared as osmotic agents for peritoneal dialysis. Kidney Int 1984;25: Volume 2 Number