Experimental Validation of Computational Model for Circulating Albumin Dialysis and Optimization of Conditions

Abstract

Traditional dialysis saves lives by removing water soluble toxins along a concentration gradient across a semi-permeable membrane. Protein bound toxins (PBTs) cannot be removed by traditional dialysis methods. By design, dialyzer pores do not allow carrier proteins like albumin or ceruloplasmin to cross the membrane. Many medical conditions such as liver failure and cholestasis are defined in part by an excess of PBTs such as bilirubin, cholic acid, copper, or manganese. End stage renal disease (ESRD) also involves the accumulation of protein bound uremic toxins (PBUTs) such as indoxyl sulfate. Removing protein bound toxins would prevent the progression of these and other diseases. PBT removal may allow recovery of the patient by reducing metabolic stress on cells in the failing organ (for example, hepatocytes in the liver). Alternatively, it may bridge the patient to a destination therapy, such as liver transplant. Finally, for ESRD patients, adding a protein bound toxin removal mechanism may improve quality of life when ESRD is used as a destination therapy. Devices for removing PBTs are commonly called artificial liver support systems, because this is the original intended use case. However, they have also been tested for treating cholestasis, poisonings, and chronic kidney disease (CKD). These devices use one or more of three approaches: The first approach is a dialysate binder suspension. A material that binds the PBT of interest is added to the dialysate. The material, referred to as the binder, has an affinity for the toxin of interest. The toxin of interest crosses the dialyzer membrane, moving along a chemical gradient. It then binds to the binder molecule. The second approach is a sorbent column to remove PBTs. Here, grains of sorbent are fixed in a plastic housing. Toxin-laden blood or plasma flows directly past the sorbent column. Toxins are removed by binding to sites on the column. The third approach is to exchange toxin laden blood, or fractions of blood, for a healthy donor blood product. Existing devices have mixed records in clinical trials. Our group recently published promising clinical data describing Acute-on-Chronic Liver Failure (AoCLF) treatment using our Advanced Multi-Organ Replacement System (AMOR). AMOR uses a human serum albumin (HSA) binder to remove toxins from blood and a charcoal sorbent column to regenerate the HSA dialysate. This is then followed by a traditional hemodialysis system which permits the removal of excess fluid and water soluble toxins, along with pH control. Unlike previous systems, AMOR removed large quantities of excess fluid (edema) from previously hypotensive liver failure patients who were refractory to fluid removal by dialysis. To our knowledge, this is the first system to achieve notable ultrafiltration in liver dialysis. To improve treatment with AMOR, or any other binder dialysis system, we developed and validated a computational model to predict PBT removal by binder dialysis. In this work we demonstrated its ability to accurately predict the final toxin concentration in benchtop trials with differing dialyzers, binder dialysate compositions, and flow rates. Assays were designed for bilirubin, albumin, creatinine, cholic acid, indoxyl sulfate, manganese, and copper. These toxins are significant for hepatic failure, ESRD, inborn errors of metabolism, cholestasis, intestinal stagnant loop syndrome, and acute poisonings. Albumin was measured to study methods to prevent albumin loss. Albumin could be lost due to binding to a sorbent column, or by transfer through a ruptured dialysis membrane. Trials were conducted to verify that albumin dialysate removes the toxins of interest in a benchtop model. The impact of flow rate on protein bound toxin removal was measured and fit using a model in which the membrane transfer coefficient of unbound toxin declines linearly with flow rate. The albumin dialysate trials were compared to negative control trials where dialysate did not contain binders. Then, the sensitivity of the system to dialysate flow rate was tested. Finally, it was shown that bovine serum albumin (BSA) at high concentration (20 g/dL) can remove bilirubin, cholic acid, and indoxyl sulfate from a human serum albumin (HSA) blood analog solution more effectively than 2 g/dL HSA or a lower (2 g/dL) BSA concentration. Charcoal column perfusion trials were done to test the ability of the FDA-approved Adsorba 300 column to remove bilirubin. A regeneration protocol was tested to determine if column saturation can be reversed by flowing dialysate through the column. Regeneration restored the rate of toxin concentration decline to the initial rate. Future study will focus on optimizing this process, validating this conclusion, and minimizing albumin loss. Dextrans were identified as promising alternatives to albumin for removing some PBTs. Dextrans are low cost compared to human serum albumin. Dextran 70 removed more bilirubin than dialysate with no added binder. Our pilot study suggests that anionic dextran sulphate likely removes copper. Dextran binder dialysis may offer a new, cost-effective method to remove PBTs. Albumins from different species offer another source of novel binders to optimize binder dialysis. Porcine Serum Albumin (PSA) and BSA dialysate both removed more cholic acid than HSA dialysate from an HSA blood analog solution. For other toxins, PSA and BSA dialysates were indistinguishable from HSA regarding the amount of toxin removed. Finally, our data suggests that the polysulfone membrane used in the Fresenius F3 dialyzer absorbs bilirubin. The quantity of bilirubin absorbed at equilibrium is measured. Better characterizing this phenomenon will enable the design of a new device for bilirubin removal using a low-risk FDA approved material. This is a critical area for further study. This includes membrane absorption of other PBTs, polysulfone rinsing and regeneration methods, and improved computational modeling of PBT removal that incorporates membrane binding.