428817 Removal of Protein-Bound Uremic Toxins Using Displacer Infusion in Hemodialysis: A Novel Mathematical Model and Its Use for Optimizing Displacer Infusion

Wednesday, November 11, 2015
Exhibit Hall 1 (Salt Palace Convention Center)
Vaibhav Maheshwari1, Stephan Thijssen1, Doris Fuertinger1, Franz Kappel2 and Peter Kotanko1, (1)Renal Research Institute, Fresenius Medical Care North America, New York City, NY, (2)Institute for Mathematics, University of Graz, Graz, Austria

Background: Hemodialysis (HD), fundamentally based on chemical engineering principles, has sustained the lives of millions of end-stage renal disease patients. HD is a membrane-based separation process where toxin-laden blood passes through a hollow fiber bundle contained in the dialyzer cartridge. In the dialyzer, based on size and charge, toxins and other solutes can diffuse between blood and the counter-current dialysate stream. Owing to the size-selectivity of HD membranes, they retain albumin on the blood side. Consequently, since many protein-bound uremic toxins exhibit a relatively high degree albumin binding, their removal is severely limited by their typically low free fractions and, therefore, low diffusion gradients. Prototypical protein-bound uremic toxins have been associated with cardiomyopathy, poor patient outcome, and numerous deleterious effects in HD patients. We have previously proposed a novel method to improve the removal of these substances via infusion of binding competitors into the arterial blood line of the extracorporeal circuit (US patent US 8,419,943B2).[1]

Methods: We developed a multi-compartment patient model and a dialyzer model depicting spatiotemporal dynamics. We chose indoxyl sulfate (IS) as a prototypical protein-bound uremic toxin and ibuprofen as the displacer substance, which is a widely-used non-steroidal anti-inflammatory drug available for intravenous infusion. Ibuprofen and IS compete for the same binding site on human serum albumin, but ibuprofen has a significantly higher association constant (1.76×105 M-1) than IS (2.3×104 M-1).[2] We modeled the IS removal during 4-hr HD session and also after dialysis with the following parameters: plasma flow rate 250 mL/min, IS and ibuprofen dialyzer clearances 150 mL/min (similar size), ultrafiltration rate 750 mL/h, initial total IS concentration 100 µmol/L, initial free fraction of IS 8%, and ibuprofen plasma half-life 2 hrs. We modeled this first without, then with infusion of 800 mg/200 mL ibuprofen (Cdisplacer) into the arterial blood line at a constant rate of 50 mL/hr (Qdisplacer). We further optimized the displacer infusion profile as a multi-objective optimization (MOO) problem where maximizing the toxin removal and minimizing the displacer infusion volume are contradictory objectives with the constraint that free plasma IS concentration should never rise above its initial level. Other cases for second objective were also considered namely, minimizing the end-dialysis plasma displacer concentration, or minimizing peak plasma displacer concentration. The formulated MOO problems were solved using non-dominated sorted genetic algorithm II. The chosen HD treatment parameters are commonly used in clinical practice of conventional HD. The protein-toxin-displacer system is assumed to follow law-of-mass-action kinetics.

Results: Total IS removal during HD without displacer infusion was 437 µmoles (plasma reduction ratio 34%), which conforms to literature data.[3] With constant infusion of ibuprofen, total IS removal increased by ~11% to 487 µmoles (plasma reduction ratio 53%). Finally, the MOO approach revealed several optimal displacer infusion profiles, which forms Pareto-optimal front. For one of the selected infusion profiles from maximized removal vs. minimized infusion volume Pareto-front, the IS removal improved to 520 µmoles (~20% improvement from baseline) with a concurrent displacer infusion volume reduction of ~14%.

Conclusion: The fundamental principle behind the removal of protein-bound uremic toxins via displacer infusion is a simple chemical reaction. The model developed here provides important insights into protein-bound toxin kinetics as well as displacer kinetics. Results from this model are in agreement with literature data. This model is expected to serve as a reliable tool for quantifying the effect of various displacer substances (alone or in combination) on protein-bound toxin removal, as well as for identification of optimal displacer infusion profiles. Displacer infusion holds the promise of substantially increased removal of protein-bound uremic. Given the well-known deleterious effects of these toxins, this should translate into improved clinical outcomes in HD patients.


  1. Kotanko, P. and N.W. Levin, Method of removing protein-bound deleterious substances during extracorporeal renal replacement treatment, 2012, Google Patents.
  2. Lin, J.H., D.M. Cocchetto, and D.E. Duggan, Protein binding as a primary determinant of the clinical pharmacokinetic properties of non-steroidal anti-inflammatory drugs. Clinical pharmacokinetics, 1987. 12(6): p. 402-432.
  3. Niwa, T., Removal of protein-bound uraemic toxins by haemodialysis. Blood purification, 2012. 35: p. 20-25.

Extended Abstract: File Uploaded