Glucose is a small molecule (180.18 g/mol) with a high diffusivity; thus, at a normal fasting plasma concentration in the range of 80–100 mg/dl (4.5–5.5 mmol/L), considerable amounts can be removed during hemodialysis using glucose-free dialysate. To prevent uncontrolled losses glucose is added to most dialysates at concentrations between 100 and 200 mg/dl (5.5 and 11.1 mmol/L).1–4 Historically, dialysate glucose concentrations as high as 1,600 mg/dl (88.8 mmol/L) have been used to support ultrafiltration across low-flux dialysis membranes essentially for the same reason glucose is used in peritoneal dialysis. However, when dialysate glucose exceeds plasma levels, the glucose gradient reverses and glucose is delivered to the patient. In fact, apart from base such as bicarbonate, glucose is the only metabolite for which the gradient is often routinely reversed. But excessive delivery of glucose may cause hyperglycemia and can be harmful by itself.1,5,6 This appears to be especially relevant for hemodialysis as the administration of glucose through the dialysate bypasses the gastrointestinal tract without formation of gastrointestinal hormones, so-called incretins, so that the insulin response is blunted and hyperglycemia is likely to be prolonged or increased.7 On the other hand, glucose-free dialysate improves potassium removal during hemodialysis8 but bears a considerable risk for hypoglycemia especially in patients with diabetes.3,9,10 Thus, the benefits of different levels of glucose in dialysate have been debated.3,9–12 However, the increasing number of patients suffering from diabetes13–15 or becoming diabetic or prediabetic being on dialysis has once more brought this question to the attention of nephrologists.16
Not much information is available regarding the transport of glucose in the dialyzer and in extracorporeal applications. A large fraction of the volume flow passing the dialyzer is occupied by red blood cells, and the mass of glucose delivered to the patient (or removed from the patient) considerably depends on the distribution volume of glucose and the diffusion of glucose into (or out of) red blood cells as they pass the dialyzer. Similar considerations apply to the calculation of clearance. Therefore, the question arises as to which degree red blood cells contribute to glucose transport in hemodialysis. Such information is also important for the correct interpretation of glucose challenge tests done during extracorporeal blood treatment.17,18
Given the situation that glucose can be delivered to the patient using a high glucose dialysate as well as removed from the patient using a glucose-free dialysate, it was the purpose of this study to determine the distribution volume of glucose in whole blood on its passage through a commercial dialyzer for both glucose delivery and glucose removal in a series of in vitro experiments.
Materials and Methods
Studies were done using regular dialysis equipment and whole bovine blood, which was anticoagulated with ethylenediamine tetraacetic acid disodium salt, dihydrate (EDTA-Na) at a concentration of 1 mg/ml. Investigations took place within 24 h of blood collection. For studies 5–10 L of blood were kept at 37°C using a water bath (Haake W26 Thermo Electron, Karlsruhe, Germany) and gently mixed with a blade stirrer (RW16 basic; KIKA Labortechnik, Staufen, Germany).
To study the delivery (henceforth termed “loading”) as well as the removal (henceforth termed “unloading”) of glucose under comparable treatment conditions two extracorporeal circulations operated by independent dialysis machines (5008H and 5008B; Fresenius Medical Care, Bad Homburg, Germany) were connected to the same blood reservoir using identical dialyzers (FX8; Fresenius Medical Care) and treatment settings, except for high glucose dialysate concentration at 11.1 mmol/L for loading, and glucose-free dialysate for unloading conditions, respectively (Figure 1). Blood inlets to extracorporeal circulations were separated from blood outlets to minimize recirculation of cleared blood. Blood entering extracorporeal blood lines was filtered (Pall SQ40S-filter; Pall Medical, Portsmouth, UK) to avoid potential clogging of individual dialyzer fibers. Ca2+ free dialysate to prevent clotting was prepared from commercial concentrates (SW405A, B. Braun, Melsungen, Germany and HämoRenol A295G, Laboratorium Dr. G. Bichsel AG, Interlaken, Switzerland) and delivered at a temperature of 37°C. All studies were done without ultrafiltration with extracorporeal blood flows set at 200, 300, and 400 ml/min. Dialysate flow was set at 500 ml/min. The actual blood and dialysate flows delivered were measured by timed volume collections.
For a given combination of experimental conditions samples were taken at the blood (bin, bout) and dialysate (din, dout) inlets and outlets of the dialyzer. A glucose concentration in the reservoir close to the steady state concentration (Equation 10, see below) was established at the beginning of each study.
Hematocrit was measured by microhematocrit technique centrifuge (Mikro 20; Hettich, Tuttlingen, Germany). Plasma viscosity at a shear rate of 100/s was measured at 37°C by a capillary rheometer (OCR-D; A. Paar KG, Graz, Austria) described earlier.19,20 Glucose concentration was measured by enzymatic reaction and electrochemical detection using the cobas b221 analyzer (Roche Diagnostics, Graz, Austria). All concentrations were measured in duplicate. Concentrations measured in whole blood refer to plasma concentrations (the red blood cells are not hemolyzed) and are therefore corrected for a constant plasma water content of 93%.
The distribution volume flow rate for glucose (also known as the effective diffusion volume flow rate) Qe was determined from solute mass balance across the dialyzer. Without ultrafiltration the relationship derived elsewhere21 simplifies to
where Qd is the dialysate flow rate, b and d refer to glucose concentrations in blood and dialysate in the inflow (index in) and in the outflow (index out), respectively.
Dialyzer solute clearance Kd follows as
The value of Kd together with Qe is used to derive the dialyzer mass transport area coefficient K0A using a modification of the Michael equation22 to account for incomplete solute equilibration throughout the blood compartment21
For a water soluble substance and incomplete equilibration throughout red blood cell water the effective diffusion volume flow rate Qe can be assumed to consist of an extracellular component, i.e., the plasma water flow rate Qpw, and a fraction α21 of the intracellular component, the erythrocyte water flow rate Qew so that
In these equations Qb refers to whole blood flow, H to hematocrit, and fpw and few to plasma and erythrocyte water fractions of 0.93 and 0.72, respectively.
The fraction α of erythrocyte water equilibrated with the solute of interest is therefore given as
If fpw, few, α, and H are known, the fraction of blood flow accessible to diffusive transport is obtained from Equations 4, 5, and 6 so that
Steady State Concentration
With clearances K1 and K2 of two parallel dialyzers operating with different dialysate concentrations d1 and d2 the change in solute concentration c for the single-pool constant volume model is given as
The steady state concentration css where dc/dt = 0 is therefore given as
and for the special case that K1 = K2
that is, for equal clearances the steady state concentration is the average of the dialysate concentrations.
The relationship between different variables as well as the relationship between identical variables obtained in subsequent measurements was examined by linear regression analysis as well as by ANOVA with Scheffé correction for multiple comparisons. Differences were assessed by t-test; p < 0.05 was considered significant to reject the null hypothesis. Unless otherwise indicated data are presented as mean ± standard deviation (SD).
The effective diffusion volume flow rate Qe for glucose was determined in five studies using whole bovine blood from different donor animals. The average hematocrit was 0.41 (range 0.36–0.50) and the average plasma viscosity was 1.39 (range 1.27–1.50) mPa·s. In each study, the mass balance across the same dialyzer was examined at three different blood flows in duplicate measurements both for glucose loading (cd = 11.1 mmol/L) and unloading (cd = 0 mmol/L) conditions. In one study measurements for one flow were repeated so that 32 measurements were available for each mode and that 64 measurements were available for final analysis. The experimental data are summarized in Table 1.
Using two dialysis machines and a parallel setup of extracorporeal circuits the same blood flow Qb and plasma water flow Qpw (Equation 5) were delivered to both dialyzers (Table 1). Despite identical machine settings (Qd = 500 ml/min) dialysate flow Qd was slightly larger (by 9.2 ± 3.1 ml/min, p < 0.001) under unloading conditions; however, dialyzer Kd (Figure 2, top panel) and effective diffusion volume flow rate Qe for glucose (Figure 2, bottom panel) were not different between treatment modes (p = ns) (Figure 2, bottom panel). On average Qe was slightly larger than the corresponding plasma water flow rate Qpw (p < 0.05, paired t-test). The ratio Qe/Qpw was not different from 1, and the fraction α (Equation 7) was not different from zero but slightly different between loading and unloading conditions (Table 1, Figure 3 top and mid panels, respectively). The dialyzer K0A (Equation 3) for glucose was 301.6 ± 45.2 ml/min and not different between treatment modes (Figure 3, bottom panel). The arterial concentration entering the dialyzer was close to the steady state concentration predicted from Equation 10.
In this study, the transport of glucose in the dialyzer under conditions resembling the in vivo situation was studied using bovine blood for the perfusion of the blood compartment. Because a major fraction of the dialyzer blood compartment is occupied by red blood cells, the transport of glucose between blood and dialysate can be expected to depend on the volume fraction of red blood cells, the hematocrit, as well as on the resistance of glucose to diffuse not only across the dialyzer but also across the red blood cell membrane. It was observed that the distribution volume for glucose in bovine blood passing the blood compartment was essentially identical to the plasma water volume both when glucose was either delivered (loading conditions) or removed from the blood stream (unloading conditions) with the use of high glucose or glucose-free dialysate, respectively. Thus, the intracellular fraction of blood water was inaccessible to glucose under both loading and unloading conditions. In other words, glucose was essentially cleared from plasma or loaded into plasma only. The mass transfer area coefficient K0A calculated from Equation 3 under consideration of the proper distribution volume flow rate Qe was not different between glucose loading and unloading conditions. Thus, if the K0A for glucose of a dialyzer is known, the clearance of glucose can be calculated from actual dialysate and plasma water flows. The plasma water flow is obtained from the whole blood flow Qb, the hematocrit H, and the fraction of plasma water fpw, as described in Equation 5. As a consequence the clearance of glucose is expected to decrease with increasing hematocrit. This dependence is strongest with solutes distributing in plasma water only.
The exclusion of red blood cell water from the exchange of glucose across the dialyzer membrane is not unexpected.23,24 The degree of glucose equilibration with intracellular water (α) can be estimated from the rate constant for glucose equilibration in whole blood (a) and from the mean transit time (τ) of blood passing the dialyzer blood compartment Vd as α = 1–exp(–a τ).25 With Vd = 74 ml according to the FX8 data sheet and Qb = 200–400 ml/min, τ = 0.370–0.185 min. The specific rate constant ks for glucose equilibration across human erythrocytes at 37°C has been determined as 0.168/min.24 With a mean hematocrit of 41% used in this study the rate constant for glucose equilibration in whole blood would be determined as a = 0.359/min.25 With this information the expected degree of glucose equilibration α is in the range between 6 and 12%. Therefore, the α obtained for bovine erythrocytes in this study (4.4 ± 20%) is plausible and comparable to that predicted for human erythrocytes.
For a small neutral solute such as glucose, the K0A should be independent of the direction of flux with either loading or unloading conditions. This was confirmed in this study (Figure 3, bottom panel).
A comparison of mass area transfer coefficients for the FX8 dialyzer shows that the K0A for glucose (301 ± 45 ml/min) determined in this study was much smaller than the K0A obtained for creatinine (589 ml/min) using clearance and blood flow data tabulated in the FX8 data sheet provided by the manufacturer. A smaller K0A for glucose is expected because glucose has a higher molecular mass (180.16 g/mol) than creatinine (113.12 g/mol).26 Estimated from the Renkin equation the diffusivity of glucose in dilute aqueous solutions at 37°C is 0.98 × 10−6 cm2/s.27 This is about 20% smaller than the estimated diffusivity of creatinine (1.21 × 10−6 cm2/s). Indeed, the in vitroK0A for glucose for the same FX8 dialyzer in aqueous solutions was determined as 432.5 ml/min, which is 26% smaller than that of creatinine.28 In addition, the diffusivity of solutes depends on the viscosity of the solution.26 The plasma viscosity was 1.39 mPa·s in this study and almost twice as high as the viscosity of water and dialysate (0.69 mPa·s) used in typical in vitro studies to determine the dialyzer K0A. It is therefore plausible to assume that the diffusivity of solutes in the blood compartment is further reduced under in vivo conditions and in studies using whole blood instead of water and saline, and that the K0A for a solute determined under such conditions is therefore lower than that accounting for differences in molecular mass only. The exact magnitude of that reduction, however, remains to be analyzed.
In conclusion, the distribution volume of glucose in blood passing the dialyzer appears to correspond to plasma water volume without measurable equilibration with red blood cell water during that transit. The diffusion volume flow rate for glucose in whole blood therefore corresponds to plasma water flow rate. Furthermore, if the K0A for glucose is known for a given dialyzer, the clearance of glucose can be predicted from actual dialysate and plasma water flows, where the latter is easily obtained from whole blood flow and hematocrit.
This study was conducted as an FFG bridge project with industrial support from Roche Diagnostics Graz GmbH. The authors thank the Austrian Research Promotion Agency (FFG) and Roche Diagnostics Graz for support of the study.
1. Sharma R, Rosner MH. Glucose in the dialysate: Historical perspective and possible implications? Hemodial Int. 2008;12:221–226
2. Dukkipati R, Kalantar-Zadeh K, Kopple JD. Is there a role for intradialytic parenteral nutrition? A review of the evidence. Am J Kidney Dis. 2010;55:352–364
3. Takahashi A, Kubota T, Shibahara N, et al. The mechanism of hypoglycemia caused by hemodialysis. Clin Nephrol. 2004;62:362–368
4. Smolle KH, Kaufmann P, Holzer H, Druml W. Intradialytic parenteral nutrition in malnourished patients on chronic haemodialysis therapy. Nephrol Dial Transplant. 1995;10:1411–1416
5. Deacon CF. Incretin-based treatment of type 2 diabetes: Glucagon-like peptide-1 receptor agonists and dipeptidyl peptidase-4 inhibitors. Diabetes Obes Metab. 2007;9(suppl 1):23–31
6. Schetz M, Vanhorebeek I, Wouters PJ, Wilmer A, Van denBerghe G. Tight blood glucose control is renoprotective in critically ill patients. J Am Soc Nephrol. 2008;19:571–578
7. Czabak-Garbacz R, Schneditz D, Zierler E, et al. Blunted insulinemia using high dialysate glucose concentration during hemodialysis. ASAIO J. 2011;57:444–450
8. Zehnder C, Gutzwiller JP, Huber A, Schindler C, Schneditz D. Low-potassium and glucose-free dialysis maintains urea but enhances potassium removal. Nephrol Dial Transplant. 2001;16:78–84
9. Jackson MA, Holland MR, Nicholas J, Lodwick R, Forster D, Macdonald IA. Hemodialysis-induced hypoglycemia in diabetic patients. Clin Nephrol. 2000;54:30–34
10. Burmeister JE, Scapini A, da RosaMiltersteiner D, da Costa MG, Campos BM. Glucose-added dialysis fluid prevents asymptomatic hypoglycaemia in regular haemodialysis. Nephrol Dial Transplant. 2007;22:1184–1189
11. Raimann JG, Kruse A, Thijssen S, et al. Fatigue in hemodialysis patients with and without diabetes: Results from a randomized controlled trial of two glucose-containing dialysates. Diabetes Care. 2010;33:e121
12. Schneditz D. Glucose-added dialysis fluid prevents asymptomatic hypoglycaemia in regular haemodialysis. Nephrol Dial Transplant. 2008;23:1066–7; author reply 1067
13. Bouffard Y, Tissot S, Delafosse B, et al. Metabolic effects of hemodialysis with and without glucose in the dialysate. Kidney Int. 1993;43:1086–1090
14. Mahnensmith RL, Zorzanello M, Hsu YH, Williams ME. A quality improvement model for optimizing care of the diabetic end-stage renal disease patient. Semin Dial. 2010;23:206–213
15. Kovesdy CP, Park JC, Kalantar-Zadeh K. Glycemic control and burnt-out diabetes in ESRD. Semin Dial. 2010;23:148–156
16. Simic-Ogrizovic S, Backus G, Mayer A, Vienken J, Djukanovic L, Kleophas W. The influence of different glucose concentrations in haemodialysis solutions on metabolism and blood pressure stability in diabetic patients. Int J Artif Organs. 2001;24:863–869
17. Schneditz D, Hafner-Giessauf H, Holzer H, Thomaseth K. Intracorporeal glucose disposal during hemodialysis after a standardized glucose load. ASAIO J. 2010;56:204–209
18. Schneditz D, Hafner-Giessauf H, Thomaseth K, Bachler I, Obermayer-Pietsch B, Holzer H. Insulinogenic index in non-diabetics during haemodialysis. Nephrol Dial Transplant. 2010;25:3365–3372
19. Schneditz D, Rainer F, Kenner T. Viscoelastic properties of whole blood. Influence of fast sedimenting red blood cell aggregates. Biorheology. 1987;24:13–22
20. Metry G, Adhikarla R, Schneditz D, Ronco C, Levin NW. Effect of changes in the intravascular volume during hemodialysis on blood viscoelasticity. Indian J Nephrol. 2011;21:95–100
21. Gotch FA, Panlilio F, Sergeyeva O, et al. Effective diffusion volume flow rates (Qe) for urea, creatinine, and inorganic phosphorous (Qeu, Qecr, QeiP) during hemodialysis. Semin Dial. 2003;16:474–476
22. Michaels AS. Operating parameters and performance criteria for hemodialyzers and other membrane-separation devices. Trans Am Soc Artif Intern Organs. 1966;12:387–392
23. Whittam R Transport and Diffusion in Red Blood Cells.. 1964 Baltimore Williams and Wilkins
24. Colton CK, Merrill EW, Reece JM, Smith KA. Diffusion of organic solutes in stagnant plasma and red cell suspensions. Chem Eng Prog Symp. 1970;99:85–100
25. Schneditz D, Yang Y, Christopoulos G, Kellner J. Rate of creatinine equilibration in whole blood. Hemodial Int. 2009;13:215–221
26. Fournier RL Basic Transport Phenomena in Biomedical Engineering.. 2007 New York Taylor & Francis
27. Renkin EM, Curry FETransport of water and solutes across capillary endothelium, in. Giebisch G, Tosteson DC, Ussing HH (eds) Membrane Transport in Biology.. 1979 New York Springer Verlag:1–45
28. Moik CClearance and mass transfer coefficient for glucose in the aqueous system (Dialysator Clearance und Massentransportkoeffizient von Glucose im wässrigen System) [Bachelor Thesis].. Graz: Graz University of Technology, 2009. Available at: http://lamp.tu-graz.ac.at/%7Eimt/bibliothek/