Continuous hemofiltration model using porcine blood for comparing filter life

Abstract
The purpose of the present study was to establish a continuous hemofiltration model using porcine blood to compare filter life. Continuous hemofiltration (CHF) experiments were performed using an in vitro hemofilter evaluation system utilizing porcine blood containing trisodium citrate in addition to nafamostat mesilate as anticoagulants. The lifetime of the hemofilter was evaluated using the transmembrane pressure and the pressure drop across the hemofilter at varying trisodium citrate concentrations. The porcine blood used in this experiment was considered to be in a slightly hypercoagulable state because of the continuous contact with non-biological materials and calcium inflow from substitution fluid. Blood containing 7 or 8 mM of trisodium citrate and nafamostat mesilate could be effectively used to compare the lifetimes of hemofilters utilized under the same conditions. In this CHF model using porcine blood, the plugging of the hollow fibers occurred shortly after the plugging of the membrane pores. In conclusion, a CHF model using porcine blood can be established by adjusting the concentration of trisodium citrate added to the blood.

Introduction
Since continuous hemodiafiltration using a polymethymeth- acrylate (PMMA) membrane can effectively and continu- ously remove various cytokines from the circulating blood by adsorption, PMMA membranes are effective for the treat- ment of severe sepsis and septic shock in patients with high blood levels of cytokines [1, 2]. PMMA membranes can adsorb large amounts of cytokines, but their membrane pores are likely to become plugged and narrowed, which may subsequently lead to a shortened lifetime of the hemofilter. However, the lifetime also depends on the patients’ clinical conditions [3, 4] and the amount of anticoagulant used. If the filter life will be evaluated under the same condi- tion, the lifetimes of different hemofilters can be compared and used as a basic performance indicator when selecting filters for clinical use. Such evaluations would also be useful for developing new filters with improved antithrombogenic- ity. These evaluations would require a continuous hemofil- tration model that utilizes animal blood to evaluate filter life. For in vitro evaluations of actual hemofilters used in clini- cal applications, several liters of experimental solutions are required. For this purpose, commercially available bovine or porcine blood obtained from a slaughterhouse is relatively easy to use. The purchased animal blood, however, might already have been activated during blood collection at the slaughterhouse followed by its transfer to the laboratory. In the present study, we first evaluated the coagulability and the levels of coagulation and fibrinolysis markers in purchased blood, compared with fresh porcine blood.

Then, continuous hemofiltration (CHF) experiments were performed using an in vitro hemofilter evaluation system uti- lizing porcine blood containing trisodium citrate in addition to nafamostat mesilate as anticoagulants. In clinical situa- tions, only nafamostat mesilate is enough to be used as an anticoagulant because blood comes in contact with non-bio- logical materials only during extracorporeal circulation. In in vitro experimental systems, however, the blood is always in contact with foreign materials; thus, larger amounts of anticoagulant or different types of anticoagulants may be required. In the present study, we used trisodium citrate in addition to nafamostat mesilate and determined the optimal trisodium citrate concentration for a continuous hemofiltra- tion model using porcine blood to evaluate the lifetimes of hemofilters.The coagulant activity and markers of activated coagula- tion and fibrinolysis were compared between purchased and fresh porcine blood to confirm that purchased porcine blood is suitable for evaluating the lifetimes of hemofilters. The experimental protocol used to obtain fresh porcine blood was approved by the Animal Research Committee of the Kitasato University School of Medicine (permit number: 2014-140). Fresh porcine blood was obtained from pigs that had been treated with clopidogrel (150 mg/day) and aspirin (300 mg/day) 1 day before and on the day of the blood collection procedure.

For blood collection, the animals were sedated and anesthetized using an intramuscular injec- tion of 2.3–8.4 mg/kg of xylazine (Xylazine inj. 2% Fujita; Fujita Pharmaceutical Co., Ltd., Tokyo, Japan), 0.5 mg of atropine (Atropine Sulfate inj. 0.5 mg “Tanabe”; Mitsubishi Tanabe Pharma Co., Osaka, Japan), and 10.2–35.6 mg/kg of ketamine (Ketalar®; Daiichi Sankyo Propharma Co., Ltd., Tokyo, Japan), followed by a 10-mg/kg/h continuous infu- sion of 2% propofol (2% Propofol inj. Maruishi;, Maruishi Pharmaceutical Co., Ltd., Osaka, Japan); the animals were then intubated and ventilated (FiO2, 35%). An 8-F catheter was introduced into the left carotid artery, and the fresh por- cine blood samples were collected.The porcine blood used for this experimental model wascollected at a slaughterhouse, and purchased from a dealer of animal blood and organs for research use (Tokyo Shiba- ura Zouki, Tokyo Japan) on the morning of the day of the experiments. Trisodium citrate was added to the blood (final concentration: 10 mM) after blood collection (no antiplateletdrug was used), and the blood was carefully transferred in a cooler to our laboratory. Penicillin and streptomycin (GIBCO, Grand Island, NY) were added to the blood at final concentrations of 100 units/mL and 100 µg/mL, respectively, and the purchased porcine blood samples were collected.Immediately after blood collection, the sample blood was transferred to a vacuum blood collection tube containing trisodium citrate (3.2%) and was centrifuged at 2500×g at 4 °C for 10 min. The plasma was then collected from the sample. The antithrombin III level was measured using a chromogenic synthetic substrate assay, the total plasmino- gen activator inhibitor 1 (total PAI-1) level was measured using a latex photometric immunoassay (LPIA), the throm- bin–antithrombin III complex level was measured using an enzyme immunoassay (EIA), and the D-dimer level was measured using latex immunonephelometry (SRL Inc., Tokyo, Japan).

Preparation and procedure for CHF experimentsThe continuous hemofiltration experiments (CHFs) were performed using purchased porcine blood. Before the CHF experiments, the hemofilter (Hemofeel® CH-1.0N; Toray Co., Ltd., Tokyo, Japan) and blood circuit (U-520, SZ-M; Junken Medical Co., Ltd., Tokyo, Japan) with both the blood inlet and outlet connected to a soft bag (FCB-3; Asahi Kasei Medical Co., Ltd., Tokyo, Japan) were primed with saline. After the saline was replaced with the blood, the blood was allowed to flow into the soft bag until the total volume of blood inside the circuit, hemofilter, and soft bag reached 1 L (0.8 L in the soft bag and 0.2 L in the blood circuit and hemofilter). The soft bag was then placed on a plate shaker (RM-300; As One Co., Osaka, Japan), covered with a water jacket circulating water at 42 °C to maintain a constant tem- perature (37 °C), and shaken at 30 cycles/min. Then, the CHF experiment was started at a blood flow rate (QB) of 100 mL/min and at a flow rate of filtration (QF) and a sub- stitution fluid flow rate (QS) of 10 mL/min. Sublood BSG® (Fuso Pharmaceutical Industries, Osaka, Japan) was used as the substitution fluid. The CHF experiments were continued until the pressure at the arterial chamber or the transmem- brane pressure (TMP) reached 300 mmHg. The maximum experiment time was 48 h because mild or severe hemolysis was observed in several experiments after 48 h of blood circulation.Nafamostat mesilate (Coahibitor®; AY PharmaceuticalsCo., Ltd. Tokyo, Japan) dissolved in 5% glucose solution (concentration 10 mg/mL) was injected into the blood circuit from the blood side inlet; a 20-mg bolus injection was used at the start of the CHF experiment, and a 20-mg/h continu- ous injection was used during the experiment. The trisodium citrate (Wako Pure Chemical Industries, Ltd., Osaka, Japan) concentration in the circulating blood was maintained at 0,6, 7, 8, or 10 mM during the CHF using substitution fluid to which trisodium citrate (final concentration adjusted to 0, 6, 7, 8, or 10 mM) had been added.

During the CHF, blood was collected to check the hematocrit and protein concentration in the blood, and the pressures at the arterial and venous sides of the circuit and at the filtrate side were continuously measured.After the start of the CHF experiment, blood was collected from the sampling port on the arterial side of the blood cir- cuit at 0 (just after start), 1, 3, 6, 20, 30, and 48 h. The hematocrit level was measured using the microhematocrit method or an automated cell counter (MEK-6450; Nihon Kohden Co., Tokyo, Japan). The total protein concentration in the blood was measured using a refractometer (SUR-JE; Atago Co. Ltd., Tokyo, Japan).The arterial side pressure (blood side inlet of hemofilter), PA, venous side pressure PV, filtrate side pressure, PF, and TMP were recorded from the monitor of the blood purifica- tion machine (TR55X or TR525; Toray Medical Co., Ltd., Tokyo, Japan) every half hour (or every 6 min after the start of a rapid increase in pressure). The TMP was calculated using Eq. (1):15 mmHg/h can be considered the times at which the pores of the membranes start to become plugged and the times at which the hollow fibers of the hemofilter start to become plugged. The value of 15 mmHg/h was determined because the variation of the value caused by the measurement error during stable periods was within 15 mmHg/h (maximum value was 13.5 mmHg/h).If no steep increase was observed during the CHF experi- ments and neither the TMP nor the pressure drop reached 200 mmHg at 48 h, the times at which the TMP and the pressure drop reached 200 mmHg and their derivative values reached 15 mmHg/h were assumed to be 48 h. If the pres- sure at the arterial chamber or the TMP reached 300 mmHg but the TMP or the arterial side pressure had only reached 200 mmHg, the time at which the other value had reached 200 mmHg was assumed to be the time at which the experi- ment ended.The statistical significance of differences between the fresh blood and the purchased blood was determined using a Stu- dent t test. A two-way ANOVA with the Tukey post hoc test was used for multiple-group comparisons of filter life. Wil- coxon signed-rank test was used for comparisons of the time at which the membrane pores started to become plugged with the time at which the hollow fibers started to become plugged. Probability values (P values) of less than 0.05 were considered to denote statistical significance.

Results
To confirm that purchased porcine blood can be used toThe derivatives of TMP (dTMP/dt) and the pressure drop (dΔPB/dt) at varying trisodium citrate concentrations were calculated for the pressure drop measured every hour (every 6 min during a steep increase in pressure). Then, the time at which the TMP reached 200 mmHg and the time at which the dTMP/dt reached 15 mmHg/h were calculated. Furthermore, the time at which the pressure drop reached 200 mmHg and the time at which the dΔPB/dt of the pres- sure drop reached 15 mmHg/h were also calculated. These times were then compared for different trisodium citrate con- centrations. The times at which the TMP and pressure drop across the hemofilter reached 200 mmHg can be considered the times at which the pores of the membranes had become plugged and the times at which the hollow fibers of the hemofilter had become plugged, respectively. The value of 200 mmHg was determined as 2/5 of the limit of membrane pressure resistance (500 mmHg). While the times at which the dTMP/dt and pressure drop across the hemofilter reachedfibrinolysis properties of purchased porcine blood were com- pared with those of fresh blood in terms of antithrombin III activity, total plasminogen activator inhibitor 1 (total PAI- 1), thrombin–antithrombin III complex (TAT), and D-dimer levels (Tables 1, 2). The reference values of antithrombin III activity in humans range from 79 to 121%. The activities of all the fresh blood samples were within the range of refer- ence values for humans. Two of the purchased blood samples were below the range of reference values. No significant dif- ference in the activities of fresh and purchased porcine blood was observed.

The reference values for TAT and total PAI-1 are below 3.0 and 50 ng/mL, respectively. The concentra- tions of all the fresh blood samples were within the ranges of reference values. Two of the purchased blood samples have values above the reference values for PAI-1, and about 2/3 of the purchased blood samples were above the range of reference values for TAT. No significant difference in the concentration of total PAI-1 was observed between the freshTand purchased porcine blood, but the TAT concentration of the purchased blood was significantly higher than that of the fresh blood (P < 0.023). The reference values of D-dimer for humans are below 1.0 µg/mL. All the samples of both the fresh and purchased blood were within the range of reference values for humans. No significant difference in the D-dimer concentration was observed between fresh and purchased porcine blood.First, we used only nafamostat mesilate as an anticoagu- lant; a 20-mg bolus injection was performed at the start of the CHF experiment, and a 20-mg/h continuous injection was performed during the course of the experiment because nafamostat mesilate is mainly used in Japan for continu- ous blood purification therapy. Under these conditions, blood coagulation occurred at a very early time (2.1 ± 0.6 h, n = 5). The coagulation of some blood samples occurred in the bag (n = 3), while for others the coagulation occurred in the hemofilter (n = 2). This result suggested that filter life cannot be evaluated using blood containing only nafamostat mesilate, which would be closest to the usual clinical set- ting, because the blood is in continuous contact with non- biological materials in the experimental model.

Accordingly, although the experimental system became more compli- cated, we were obliged to use trisodium citrate in addition to nafamostat mesilate. Using nafamostat mesilate to inhibit the serine protease activity of coagulation factors, we attempted to determine the optimal trisodium citrate concentration for a continuous hemofiltration model using porcine blood to evaluate the lifetimes of hemofilters.During the CHF experiment, blood was collected andexamined to determine whether the hematocrit and protein concentrations in the blood were affected by the trisodium citrate concentration (Fig. 1). The hematocrit level in the blood decreased slightly but was stable during the CHF experiment, while the total protein concentration in theblood decreased with time. No significant differences were observed among the different trisodium citrate concentra- tions that were evaluated.The TMP and a derivative of the TMP (Fig. 2a, b), as well as the pressure drop across the hemofilter, ΔPB, and a derivative of the pressure drop (Fig. 2c, d), were measured during CHF experiments performed with varying trisodium citrate concentrations. The TMP and the pressure drop, as well as their derivatives, remained stable and constant for several hours and then increased steeply. The same tendency was observed for all the data, but the time at which the steep increase occurred differed according to the trisodium cit- rate concentration and individual blood samples. For almost all blood samples containing a trisodium concentration of 10 mM, a steep increase was not observed during 48 h of CHF.The times at which the TMP reached 200 mmHg and the dTMP/dt reached 15 mmHg/h were compared among the different trisodium citrate concentrations (Fig. 3a, b). Both times increased significantly as the trisodium citrate con- centration increased (P < 0.05 for 7 vs. 8 mM, P < 0.01 for the others).

The times at which the pressure drop across the hemofilter reached 200 mmHg and the d ΔPB/dt of the pres- sure drop reached 15 mmHg/h were also compared among the different trisodium citrate concentrations (Fig. 3c, d). Both times increase significantly as the trisodium citrate concentration increased (P < 0.05 for 7 vs. 8 mM, P < 0.01 for the others).The TMP and its derivative and the pressure drop and its derivative both showed the same dependence on the triso- dium citrate concentration. Consequently, we can evaluate filter life using either of these parameters: the time until almost complete plugging and the time until the start of plugging. In some cases, the TMP reached 300 mmHg and the experiment was stopped, but the pressure drop hadnot yet reached 200 mmHg. In such cases, the time when the pressure drop reached 200 mmHg was assumed to be the time when the experiment was stopped (because we cannot continue the experiment if the TMP had reached 300 mmHg). When the derivative was compared, such cases were rare, meaning that for this evaluation model, the time until the start of the plugging of pores or hollow fibers as calculated using the derivatives tended to be more accurate than the time at which the pores or fibers had become almost completely plugged.When the time at the start of the plugging of the mem- brane pores was compared with that of the start of the plug- ging of the hollow fibers (Fig. 3b, d), the time at which the membrane pores started to become plugged was significantly earlier than the time at which the hollow fibers started to become plugged at trisodium citrate concentrations of 7 (P < 0.01) and 8 mM (P < 0.05). This means that the plug- ging of the hollow fibers occurred after the plugging of the membrane pores in this CHF model.

Discussion
The main findings of the present study were as follows: (1) the purchased porcine blood was slightly coagulated at the time of examination, but the blood had retained its coagula- bility; (2) blood containing 7 or 8 mM of trisodium citrate was useful for mimicking hypercoagulable blood and was suitable for comparing the lifetimes of hemofilters; and (3) the plugging of the hollow fibers occurred shortly after the plugging of the membrane pores in the present CHF experi- ment model. The antithrombin III activity was sufficiently high in the fresh blood, but the activity was slightly decreased in some of the purchased blood and the TAT concentration was increased in about 2/3 of the purchased blood, meaning that the antithrombin III had been consumed to form TAT in some of the purchased blood. The purchased porcine blood was considered to have undergone some coagulation during the collection and transport of the blood to the laboratory. If the endothelial cells had been stimulated by endotoxin and/or thromboxane due to endothelial damage by incision during the collection of the blood, the PAI-1 level would have been elevated in the purchased blood, which may lead to the inhibition of fibrinolysis and a hypercoagulable state. However, the coagulability of most of the purchased blood samples was within the normal range for humans, mean- ing that the activation or stimulation of coagulation rarely occurred during the collection and transport of the blood. The D-dimer levels were within the normal range, so only slight coagulation and fibrinolysis had occurred in the pur- chased blood.

To consider the effects of trisodium citrate on blood coagulation in this experiment, the concentration of cal- cium is important. In the course of CHF with substitution fluid containing calcium, the total calcium levels in the blood decreased and reached 7.0 mg/dL (0 h: 7.6 ± 0.4; 1 h: 7.2 ± 0.2; 3 h: 7.1 ± 0.1; 6 h: 7.0 ± 0.5; 10 h: 7.0 ± 0.1 mg/ dL; n = 5), when the levels were measured using an enzy- matic assay with α-amylase and 2-cloro-4-nitrophenyl-4-Ο- β-D-galactopyranosylmaltoside (Diacolor liquid Ca; Toyobo Co., Ltd., Osaka, Japan). The calcium concentration of the substitution fluid was 1.75 mmol/L (7.0 mg/dL), meaning that within about 3 h, the calcium concentration of the blood and the calcium concentration of substitution fluid became almost equal. Similar results are expected for other triso- dium citrate concentrations. The difference in calcium ion concentration is thought to influence the filter life. Several of the blood samples with a trisodium citrate concentra- tion of 10 mM did not exhibit a steep increase but rather a gradual increase in the TMP and a pressure drop during the 48-h CHF experiment. The hematocrit level also gradually decreased, probably because of hemolysis and a decrease in the total protein level, including coagulation factor, because of solute removal during CHF. Since the reliability of the data was expected to decrease for longer experimental peri- ods, we stopped the experiments at 48 h. Accordingly, a trisodium citrate concentration of 10 mM is probably inap- propriate for evaluating filter life but could be used to eval- uate the effects of protein-induced membrane fouling. In almost all the samples with a trisodium citrate concentration of 6 mM, blood clots became trapped at the mesh in the soft bag filled with blood or in the chamber of the blood circuit.

Therefore, a concentration of 6 mM is probably also unsuita- ble for evaluating filter life. The time at which the membrane pores started to become plugged was significantly earlier than the time at which the hollow fibers started to become plugged at trisodium citrate concentrations of 7 and 8 mM. This model using trisodium citrate concentrations of 7 and 8 mM can be used to evaluate the lifetimes of hemofilters based on the plugging of the hollow fibers triggered by the plugging of the membrane pores. According to the method concentration or the hematocrit level to reduce the process of centrifugation and mixing, which can potentially induce platelet activation. However, the total protein concentra- tion and hematocrit levels have a direct effect on the TMP. These factors may partly explain the variation in the TMP in the present experiments and may affect the filter life as determined using this model. Third, the characteristics of the purchased blood samples, especially their coagulability, dif- fered among the samples. Trisodium citrate cannot be used to eliminate differences in coagulability because it cannot equalize the consumed coagulation factors in blood prior to the experiments. To overcome these drawbacks, two options can be selected: blood obtained from a single animal can be divided into two portions and used to compare two hemofil- ters [6], or the coagulant activity and markers of activated coagulation and fibrinolysis can be examined before the experiment and used to identify blood samples suitable for use in the experiments. While the results obtained from this model should be interpreted with consideration of the above points, purchased blood appears to be useful for evaluating the lifetimes of hemofilters on a like-for-like basis under the same experimental conditions.

Conclusions
A continuous hemofiltration model using porcine blood can be established by adjusting the concentration of trisodium citrate added to the blood. Blood samples with 7 or 8 mM of trisodium citrate plus nafamostat mesilate that are then con- tinuously circulated through non-biological materials during continuous hemofiltration are considered to be in a hyper- coagulable state and can be used to compare the lifetimes of hemofilters under the same experimental conditions.