Accounting for the Structure–Property Relationship of Hollow-Fiber Membranes in Modeling Hemodialyzer Clearance Authors: Anton Kozmai, Mikhail Poroznyy, Violetta Gil, Dmitrii Butylskii, Dmitry Lopatin, Aleksey Rodichenko, Igor Voroshilov, Artem Mareev, Victor Nikonenko, Johannes Carolus (John) Jansen Published: 2024 ABSTRACT AbstractThe relevance of the hemodialysis procedure is increasing worldwide due to the growing number of patients suffering from chronic kidney disease. Taking into account the structure of dialysis polymer membranes is an important aspect in their development to achieve the required performance of hemodialyzers. We propose a new mathematical model of mass transfer that allows hollow-fiber membrane structural parameters to be taken into account in simulating the clearance (CL) of hemodialyzers in a way that does not require difficult to achieve close approximation to the exact geometry of the membrane porous structure. The model was verified by a comparison of calculations with experimental data on CL obtained using a lab-made dialyzer as well as commercially available ones. The simulations by the model show the non-trivial behavior of the dialyzer clearance as a function of membrane porosity (fp) and the arrangement of pores (α). The analysis of this behavior allows one to consider two strategies for increasing the CL of the dialyzer by optimizing the polymer membrane structure: (1) creating a membrane with a well-structured pore system (where α → 1) since doubling α at a high enough fp can lead to an almost tenfold increase in CL; (2) increasing the porosity of the membrane characterized by a random arrangement of pores (α → 0), where, at a relatively low α, a sharp increase in CL is observed with a small increase in fp over a certain threshold value.Keywords: hemodialysis, membrane, structure–property relationship, clearance, mathematical modeling, urea, creatinine, phosphate, porosity 1. Introduction An analysis of the hemodialyzer and dialysis filter market reveals that it will reach USD 132.5 billion by 2030, with a projected compound annual growth rate of 6.7% [1]. Such market dynamics are conditioned by the increase in the number of people in need of renal replacement therapy (RRT), which is primarily due to the growth in the elderly population that suffers from chronic kidney disease (CKD) in significantly greater numbers than other age groups [2]. According to a report on CKD prevalence in the United States (2023) presented by the Centers for Disease Control and Prevention, 34% of people aged 65 years and older suffer from CKD, followed by 12% of people aged 45 to 64 years and 6% of people aged 18 to 44 years [3]. However, according to the recent “World Population Prospects” report by the United Nations [4,5], the population aged 65 and older is growing faster than other age categories. In 2020, the percentage of this age group in the general population was 10%; however, by 2050, it is expected to increase to 16%, and by the 2080s, it is predicted to exceed 20%. An important factor in the prevalence of CKD is that CKD is a common complication of pathologies such as diabetes mellitus and arterial hypertension, which are more common in older people [6,7]. Such statistics may indicate that, in the future, the need for medical services for patients with CKD will increase. In broad terms, kidney disease is characterized by the loss of the bodys ability to maintain its electrolyte balance and remove metabolic waste products. In acute renal failure, the rapid deterioration of renal function is potentially reversible with proper treatment, but the chronic course of the disease results in the progressive and irreversible loss of this function [8,9]. RRT is mandatory in end-stage CKD. Kidney transplantation may be required to treat CKD; however, it is not able to cover all the patients in need. In addition to the obvious limitation of donor resources, it should be noted that a significant number of patients with end-stage CKD cannot be subjected to kidney transplantation for various medical reasons [10,11]. Thus, in end-stage CKD, patients have to rely on extrarenal blood purification methods, such as hemodialysis and hemofiltration, as well as peritoneal dialysis. Compared to hemodialysis, hemofiltration is a more resource-intensive and technically sophisticated method. However, a number of studies [12,13] do not note significant advantages of one method over another regarding the effectiveness of therapy. Despite a considerable number of arguments in favor of peritoneal dialysis, including a better quality of life [14,15], the vast majority of people worldwide receive RRT in the form of hemodialysis [16]. Moreover, due to the functional deterioration of the peritoneal membrane, patients must switch to hemodialysis after a median of 3.7 years of peritoneal dialysis [17]. A significant trend in the global hemodialysis technology market toward engineering developments designed to improve the efficiency and effectiveness of hemodialysis procedures has been observed. In particular, attention is being paid to improving hemodialysis filters, namely in terms of membrane selectivity/permeability to metabolic waste products [18], hemocompatibility and resistance to biofouling [19,20], as well as adaptation of the filter configuration for specific tasks [21,22]. In addition to the above topics, special attention is paid to a personalized approach to ensure effective treatment since patients with end-stage CKD may have different clinical presentations that do not allow a universal treatment. Taking these aspects into account when improving filters can be key to improving patient outcomes, speeding up dialysis treatment, reducing treatment time and increasing overall patient satisfaction and comfort. Mathematical modeling (MM) can serve as an important tool for the evaluation, optimization and control of hemodialysis therapy, both for everyday clinical use and for the study of its fundamental principles. Indeed, theoretical works in the field of modeling the hemodialysis process have made a significant contribution to the development of hemodialysis technology. The use of MM has made it possible to achieve certain successes in the field of hemodialyzer design and has also laid the basis for evaluating the results of the hemodialysis process using mass transfer kinetics concepts [23]. The theory of mass transfer in a hemodialyzer was proposed in 1981 [24] and has been applied in many studies [25,26,27]. This theory works by satisfactorily applying it to the transport of low molecular weight substances. For larger molecules, a more complex theory was developed taking into account the diffusion boundary layers adjacent to the membrane [28]. In recent years, with the widespread deployment of high-level programming software, there has been significant progress in the theoretical description of the hemodialysis process. In general, a comprehensive review of current research on the modeling of the hemodialysis process for the study and optimization of dialysis therapy can be found in recent works by Pstras et al. [23] and Miyasaka et al. [29]. Works on MM, describing the kinetics of mass transfer through the membrane, have made it possible to obtain the main parameter characterizing the efficiency of the hemodialysis process, clearance [23,29], which is used, among other things, to compare the cleaning capacity of hemodialyzers. It should be noted that some existing models use a phenomenological approach to describe the transfer of solutes through the membrane and do not take into account its structural characteristics, which directly determine its physicochemical properties. However, there is another approach focused on the structural parameters of the membrane when determining its permeability. In this approach, the well-established theory of free volume has been long and successfully applied to describe transport across membrane structures [30,31,32]. This theory attributes the main role in determining the transport characteristics of membranes to the structural parameter of tortuosity. Generally, the tortuosity of porous media is defined as a function of the volume fraction of pores (porosity). However, many studies provide rather different mathematical expressions for this relationship [33,34,35,36,37]. This can be explained by the rather vague concept of tortuosity, which has multiple definitions, and, possibly, by the fact that separate scientific groups use geometrical and different physical tortuosities interchangeably [38,39]. In most cases of tortuosity determination via porosity, one drawback is the necessity to assume and set geometrical parameters (the shape and characteristic sizes) of pores. Thus, the accuracy of a model depends on how adequately and closely these hypothesized parameters approximate to the complex structure of porous media. The aim of this study is to develop a new mathematical model of solute transport in dialysis that takes into account the membrane structure–property relationship and allows for the calculation of clearance and other hemodialyzer parameters. Considering the characterization of the membrane structure, in order to determine tortuosity as a function of porosity, we introduce the modification of the microheterogeneous model, taking into account the presence of only one conducting phase in a membrane, which leads to a relatively simple mathematical apparatus. Within the framework of the model, using porosity (the volume fraction of pores) and the introduced structural parameter of pore arrangement makes knowledge of the phase element shape redundant and removes the restrictions of its a priori assumption, unlike in the abovementioned existing approaches. The model was verified using experimental data obtained with a specially designed hemodialyzer during urea, creatinine and phosphate dialysis. The adequacy of the model was confirmed by a comparison with the clearance simulation of several commercially available hemodialyzers with data provided by their manufacturers. 2. Theoretical 2.1. Model Formulation The most common design of hemodialyzers includes a cylindrical casing in which a bundle of hollow-fiber membranes is sealed. Blood is pumped through the fibers, and the dialysate solution is pumped through the compartment formed by the outer surface of the membranes and the walls of the dialyzer casing. Generally, in such a system, three layers where the transport of substances occurs can be distinguished: a membrane and two films of solution adjacent to both surfaces of the membrane. Accordingly, in this paper, the theoretical description of mass transfer through membranes in such a dialyzer is based on the following representation of the membrane system. The dialysis system under study consists of two compartments: the diluate compartment (A) and the dialysate compartment (B), which are separated by a membrane (Figure 1). Scheme of the simulated system geometry. DBL1 and DBL2 are diffusion boundary layers adjacent to the membrane at the side of compartments A and B, respectively. The arrow indicates the direction of solutes transport. Along with the components of the hydraulic system (tubing, intermediate tanks), compartments A and B have volumes VA and VB, respectively. A model mixed solution containing typical uremic toxins (urea, creatinine and phosphates) and sodium chloride at a given pH circulates through compartment A; an individual solution of sodium chloride at a given pH circulates through compartment B as well as through the intermediate tanks. Phosphoric acid was chosen to simulate phosphate transport. For the sake of simplicity, it is assumed that the concentration of solutes remains nearly constant along the length of the compartments, i.e., the concentration at any moment in time is the same in the volumes of the compartments and in the corresponding intermediate tanks. The equilibrium between H+ and OH− ions and the water molecules in the solution is described by the following equation: The reactions of urea protolysis are not taken into account in the model since the pH of the solution in the compartments of the system and in the membrane is taken to be equal to 7.4 (the average pH of human blood). At this pH value, the urea in the solution has the form of a neutrally charged molecule (pKa of urea is 0.1 [40]). However, the phosphoric acid at this pH exists in the form of singly (H2PO4−) and doubly (HPO42−) charged anions (Figure 2). The equilibrium constant of the protolysis reaction at 37 °C is [41] Distribution of the phosphoric acid species (in mole fractions) as a function of pH. Hereafter, the anions H2PO4− and HPO42− are denoted as P− and P2−, respectively; the particles CH4N2O and C4H7N3O are denoted as Ur and Crn, respectively. The one-dimensional transport of solutes (Figure 1) in the membrane and in the diffusion boundary layers (DBLs) is described by the following set of equations: –the Nernst–Planck equation the Nernst–Planck equation –the electroneutrality condition the electroneutrality condition –the condition of zero current flow the condition of zero current flow –the equation of material balance the equation of material balance where Ji is the flux density, Di is the diffusion coefficient, zi is the charge and Ci is the concentration of solute i (i = Ur, Crn, P−, P2−, Na+, Cl−, H+, OH−); t is the time; x is the coordinate normal to the membrane surface; β = 1 for the solution and β = ξ for the membrane; ξ is the tortuosity of the conduction pathways in the membrane (see Section 2.3.1); R is the universal gas constant; T is the absolute temperature; F is the Faraday constant. The local change in the phosphoric acid species concentration as a function of pH is expressed using Equation (3), as follows: where CPtot=CP−+CP2− is the total concentration of all phosphoric acid species. The boundary conditions imply the flux continuity condition at the membrane/solution interfaces and the concentration continuity condition at the DBLs/solution and membrane/solution boundaries where CiA and CiB are the concentration of solutes i in A and B compartments, respectively; the overbar relates to the membrane. Changes in the concentration of species in compartments A and B, as well as the local concentration in the DBLs and in the membrane, are calculated from the system of equations formed by Equation (7) taking into account Equation (2): where k1 and k−1 are the dissociation and recombination rate constants in the reaction, described by Equation (2); kd and kr are corresponding constants for water. The numerical solution of the series of Equations (4)–(6) and (12)–(19), taking into account Equations (8) and (9) and the boundary conditions (10) and (11), allows one to find the change in the local concentration of all species in the studied membrane system. At the beginning of the dialysis process (at t = 0), a uniform distribution of concentrations in the DBLs is assumed, equal to the initial concentrations of the feed solutions. For the sake of simplicity, the concentration of species in the membrane at t = 0 is linearly distributed between its left- and right-hand interfaces. The numerical solution was implemented in MATLAB (R2014a) software. The membrane, as well as the DBLs, were divided into elementary segments uniformly distributed along the spatial coordinate. For each segment of the system, the algorithm first used by Larchet et al. [42] and described in detail by Mareev et al. [43] was applied. This algorithm involves the use of an explicit Euler finite-difference scheme to approximate the derivatives. The time step was determined in such a way as to satisfy the Courant–Friedrichs–Lewy condition. 2.2. Determination of the Dialyzer Clearance We propose to use the above-described model to calculate the clearance (CL) that a specific dialyzer can provide. The equation for calculating CL reads as follows [29]: where θ=−πdhfNhflhf1−QA/QB/QA; QA and QB are the volumetric flow rates in the diluate and dialysate compartments, respectively (0≤QA