Impact of the C2/C6 Ratio of High-molecular-weight Hydroxyethyl Starch on Pharmacokinetics and Blood Coagulation in Pigs

The animal experiments were performed according to the guidelines of the Swiss Federal Veterinary Office. The protocol was approved by the Veterinary Office of the Canton de Vaud, Switzerland. The study group consisted of 30 pigs with a mean body weight of 41.2 ± 3.6 (SD) kg. The pigs were fasted overnight but allowed free access to water.

The evening before the study, the pigs were separated from their group and premedicated with 150 μg clonidine (Boehringer Ingelheim [Schweiz] GmbH, Basel, Switzerland) via  intramuscular injection. Before the transportation, a dose of 0.5 mg/kg body weight (BW) midazolam (Roche Pharma AG, Reinach, Switzerland) was administered intramuscularly. Upon arrival in the laboratory, the animals received intramuscular premedication of 1 mg/kg BW midazolam (Roche Pharma AG) and 20 mg/kg BW ketamine (Veterinaria AG, Zurich, Switzerland). Once sedation was obtained, anesthesia was induced by administration of 3% halothane (Dräger, Luebeck, Germany) by mask, followed by tracheal intubation.

Controlled ventilation was performed. An end-tidal volume of 10–13 ml/kg BW and a ventilation rate of 13–18/min were adjusted to maintain partial pressure of carbon dioxide at 35–40 mmHg before the onset of the protocol (Ventilator Dräger Sulla 909 V; Dräger). The inspired oxygen fraction was monitored and maintained at 1.0 during surgical instrumentation and decreased to 0.4 (air–oxygen mixture) during the experimental protocol. The ear vein was cannulated (18-gauge cannula). Anesthesia was now switched to a total intravenous anesthesia using 5–10 mg · kg BW⁻¹· h⁻¹propofol (B. Braun Medical AG, Emmenbruecke, Switzerland), 0.6 ml · kg⁻¹· h⁻¹midazolam (Roche Pharma [Schweiz] AG, Reinach, Switzerland), and 25 μg · kg BW⁻¹· h⁻¹fentanyl (Sintetica S.A., Mendrisio, Switzerland) during surgical instrumentation. Fentanyl dosing was lowered after the surgical instrumentation to a dose of 3–7 μg · kg BW⁻¹· h⁻¹, according to the requirements of each individual pig. NaCl 0.9% solution was infused in a dosage of 10 ml · kg⁻¹· h⁻¹during the entire period of anesthesia. Glucose 20% solution was administered according to the requirements of each individual pig to keep the blood glucose concentration in a range of 3.7–5.2 mm.

A catheter was placed into the right internal jugular vein for administration of the intravenous anesthesia. Another catheter was placed in the ipsilateral carotid artery for blood withdrawal for laboratory measurements, arterial blood gas analyses, and continuous measurement of arterial blood pressure. In the contralateral internal jugular vein, a catheter was inserted for HES administration and later on for 20% glucose administration. Further monitoring of the animals included continuous three-lead electrocardiogram, heart rate reading, and continuous body temperature monitoring by a probe inserted in the esophagus. In addition, a transmural bladder catheter was inserted performing a minilaparotomy. Sheets, a ventilator, and ice were used if necessary to keep body temperature at 38°± 1°C.

Thin-boiling waxy maize starch was suspended in water, activated by means of sodium hydroxide, and allowed to react with ethylene oxide for 2 h at 40°C. The amounts of waxy maize starch and ethylene oxide were chosen to yield HES with a molar substitution of 0.42. HES C2/C6 ratios of 2.8 and 5.6 were obtained by changing the quantities of sodium hydroxide during hydroxyethylation. The raw HES species were hydrolyzed by hydrochloric acid to a molecular weight of 650 kd, treated with activated carbon, purified by ultrafiltration, and diluted to a final concentration of 6% (wt/vol) in isotonic saline, filled in glass bottles of 500 ml each and heat-sterilized at 121°C for 20 min.

The mean weight average molecular weight (MWw) of HES was determined by gel permeation chromatography/multiangle laser light scattering (GPC-MALLS; Wyatt Technology, Woldert, Germany) at a flow rate of 1 ml/min in a 70 mm phosphate buffer pH 7.0 using serial GPC columns HEMA Bio 40, 100, and 1,000 (PSS, Mainz, Germany). Molecular weight was calculated using ASTRA Software (Wyatt Technology). The weight averaged molecular weight, calculated as MWw =Σ^I(Mⁱ²Nⁱ)/Σⁱ(MⁱNⁱ), gives higher statistical weight to larger molecules. In contrast, the number averaged molecular weight, calculated as MWn =Σ^I(MⁱNⁱ)/ΣⁱNⁱ, gives equal statistical weight to each molecule. The polydispersity index MWw/MWn describes the distribution of the molecular weights, with values near 1 for a narrow distribution.

Molar substitution was determined by gas chromatography after transforming hydroxyethyl groups into ethyl iodine by hydriodic acid in the presence of adipic acid using a gas chromatograph Perkin Elmer Autosystem (Boston, MA). The C2/C6 ratio was determined after hydrolysis of HES by sulfuric acid and gas chromatographic separation of the sialylated hydroxyethylated glucose derivatives using a gas chromatograph Carlo Erba Mega 5300 (Milano, Italy). In table 1, the physiochemical parameters of the investigated HES solutions are shown.

The animals were randomized into two groups of 15 pigs each, receiving either 6% HES 650/0.42/2.8 (test group) or 6% HES 650/0.42/5.6 (reference solution). The treatments were strictly blinded over the whole study period. Unblinding was delayed until the data analysis was finalized. After completion of all surgical preparations, the first of 13 blood samples were taken for baseline values using citrated blood containers. Directly after baseline blood sampling, the administration of the solution started. A top-load dose of 30 ml/kg BW was infused over exactly 30 min. During and after the infusion, blood withdrawals were performed at 10, 20, 30, 40, 50, 60, 90, 120, 150, 270, 390, 510, and 630 min after the start of the infusion.

Arterial blood samples were collected using heparinized syringes (BD Preset; BD Vacutainer Systems, Plymouth, United Kingdom). Hemoglobin concentration was determined immediately after collection using the Rapidlab 865 analyzer (Bayer Vital GmbH, Fernwald, Germany).

Albumin concentration was measured by means of enzyme-linked immunosorbent assay (ELISA) using the sandwich technique. ELISA kit starter accessory package E 101 (Bethyl, Montgomery, TX) was combined with the Pig Albumin ELISA Quantification Kit E100-110 (Bethyl). Briefly, coating of the wells with the capture antibody and blocking was effected before the increasingly diluted plasma samples were incubated for 60 min. These were linked with antibody conjugate for another 60 min. In a last step, enzyme-substrate solution was added and quantification was measured by means of calorimetry at 490 nm (ELX808; BioWhittaker Inc., Walkersville, MD).

Citrated whole blood was used to issue a Thrombelastogram®. An initial incubation for 1 h in a 37°C water bath was performed with the citrated whole blood samples,16followed by blood recalcification and TEG® measurements using two computerized Thrombelastography® coagulation analyzers 5000 (Haemoscope Corporation) with two channels for each analyzer. Four samples were analyzed at a time with a randomly chosen channel sequence. The following TEG® parameters are reported: reaction time (r time), coagulation time (k time), maximal amplitude, angle α, elastic shear modulus, and coagulation index.16,17 

For further laboratory measurements, blood samples were immediately centrifuged at 3,000 rpm for 15 min at 4°C for separation of plasma and blood cellular components (Rotanta/RP; Hettich, Bäch, Switzerland).

Prothrombin time and activated partial thromboplastin time (aPTT) were determined on an automated coagulation analyzer (BCS; Dade Behring, Marburg, Germany) using a prothrombin time reagent containing recombinant tissue factor (Innovin; Dade Behring) and an aPTT reagent containing ellagic acid (Actin FS; Dade Behring), respectively. Functional activity of vWF was determined in a commercial ristocetin-cofactor assay (vWF RCA; Dade Behring) on an automated coagulation analyzer (BCS). Briefly, vWF activity was assessed by the ability to agglutinate fixed human platelets in the presence of ristocetin. Agglutination was measured turbidimetrically using the coagulation analyzer. FVIII was assessed functionally using FVIII-deficient plasma according to the manufacturer’s instructions (Dade Behring).

Hydroxyethyl starch concentration was quantified after extraction from plasma and hydrolysis to glucose monomers. Briefly, plasma samples (1 ml) were incubated at 100°C for 60 min after addition of 0.5 ml KOH solution 35% (wt/wt) (Fluka, Buchs, Switzerland). HES was precipitated by adding of 10 ml ice-cold absolute ethanol (Fluka) to the supernatant of the reaction mixture and acid hydrolyzed in 2N HCl (Fluka) for 60 min at 100°C. Glucose determination was performed using an enzymatic test kit based on hexokinase/glucose 6-phosphatase (Boehringer Mannheim, Darmstadt, Germany). For determination of HES molecular weight, plasma proteins were eliminated by trichloroacetic acid precipitation (6.4% [wt/wt] end concentration), and neutralized supernatants were analyzed by GPC-MALLS at a flow rate of 1 ml/min in 70 mm phosphate buffer, pH 7.0, using serial GPC columns HEMA Bio 40, 100, and 1,000.

Pharmacokinetic parameters were obtained through individual fitting of the plasma concentration data with a biexponential model using the software Kinetica, version 4.3 (InnaPhase Corp., Philadelphia, PA). Curve fitting was performed by nonlinear regression, using weights of 1/Y^obs2to accommodate for data heteroscedasticity. The use of a two-compartment model seemed appropriate to describe the concentration curves:

where c¹= relative coefficient for the central compartment, λ¹= slope of the distribution phase, λ^z= slope of the elimination phase, V¹= volume of the central compartment, and R⁰= rate of HES administration, i.e. , the dose divided by the infusion duration. A one-compartment model showed significant misfit, whereas a three-compartment model seemed overparameterized, with high SE and strong intercorrelation values of the estimates. In addition, a visual inspection of diagnostic plots such as observed versus  calculated values (fig. 1) and residuals versus  time (not shown) did not reveal any systematic trend suggestive of model misfit. Further macroconstants (area under the curve; CL = systemic clearance = ratio of dose over area under the curve; t^1/2= terminal half-life = log2/λ^z, C^max= end-of-infusion level) and microconstants of the model (k¹², k²¹, and k^el, i.e. , transfer constants between compartments and to elimination) were derived by the Kinetica software using standard formulae.

Sample size has been determined by a power analysis based on a previous in vivo  study.13To obtain a power of 80% with an estimated difference between groups of 7.5% using TEG® parameters angle α and maximal amplitude and a SD of 5%, a total sample size of 30 pigs has been determined with a type I error of 0.05. The high-molecular-weight HES solution with the low C2/C6 ratio of 2.8 (HES 650/0.42/2.8) was compared with the high-molecular-weight HES solution with the C2/C6 ratio of 5.6 (HES 650/0.42/5.6) using the JMP 5.1 statistical package (SAS Institute, Inc, Cary, NC). Data (with and without baseline correction) were analyzed using the Shapiro-Wilk test for normality and a two-way analysis of variance for repeated measures on one way (time) with the Greenhouse-Geisser correction for assessing solution and time effects. Pharmacokinetic parameters obtained from model fitting were thereafter compared between HES 650/0.42/2.8 and HES 650/0.42/5.6 by way of the unpaired Student t  test. No data transformation was applied. Results are expressed as mean ± SD.

In total, 31 animals had to undergo the study protocol to complete the investigation in 30 because of erroneous HES infusion in one individual. There were no differences between groups at baseline (table 2).

The HES concentrations increased progressively in both groups during infusion and decreased from the end of infusion until 630 min, following a biexponential profile. There were no significant differences between the solutions (figs. 2A and B).

A two-compartment model for the calculation of the HES concentration was able to fit well the concentration curves of the solutions, producing an excellent correlation between measured and calculated values (R  ²= 0.991) (fig. 1).

Hydroxyethyl starch 650/0.42/2.8 had a 29% higher elimination constant k^elthan HES 650/0.42/5.6 (P < 0.001). Also compared with this solution, a significantly smaller area under the plasma concentration–time curve was found (19%; P = 0.026). All other pharmacokinetic parameters showed no significant differences between the groups (table 3).

In vivo , the MWw (P = 0.004) and the MWn (P = 0.009) were smaller for HES 650/0.42/2.8, compared with HES 650/0.42/5.6 (figs. 2C and D). The index of polydispersity MWw/MWn revealed no significant differences between the HES solutions (fig. 2E).

Parameters of plasma coagulation such as prothrombin time (seconds), aPTT (seconds), and functional FVIII and vWF activity (%) changed significantly over time (P < 0.001 for all), but no significant between-group differences were found. The maximum effect was reached immediately after the end of infusion at 30 min (figs. 3A–D).

All TEG® parameters showed significant time-dependent alterations after the infusion irrespective of the HES solution used (P < 0.001 for all). The compromising effect on Thrombelastography® parameters reached its maximum at the end of infusion (30 min for k time, maximal amplitude, angle α, coagulation index, and shear elastic modulus), only for r time the maximum was reached after 20 min, during the infusion. No significant between-group differences were found in TEG® (figs. 4A–F).

During the HES administration, the hemoglobin concentration decreased continuously from 8.1 ± 1 g/dl to 5.9 ± 0.6 g/dl (HES 650/0.42/5.6) at the end of the infusion (30 min), respectively, and from 8.6 ± 1.1 g/dl to 6.2 ± 0.6 g/dl (HES 650/0.42/2.8) (P < 0.001 for all). Thereafter, hemoglobin concentration increased progressively to values close to the baseline values. After 630 min, the hemoglobin was 7.2 ± 0.9 g/dl for HES (650/0.42/5.6) and 7.2 ± 0.7 g/dl for HES (650/0.42/2.8). No significant differences were found between the tested solutions (fig. 5A).

The albumin concentration showed a similar behavior with an initial decrease until the end of infusion followed by a subsequent increase without reaching the baseline values. No significant differences were found between the tested solutions (fig. 5B).

The primary aim of this study was to examine the single-dose administration of a new high-molecular-weight, low-substituted HES with a low C2/C6 ratio regarding its pharmacokinetic profile, compared with a reference solution. We found that a reduction of the C2/C6 ratio from HES 650/0.42/5.6 to HES 650/0.42/2.8 modifies to some extent the pharmacokinetic profile, resulting in lower area under the curve surface and higher elimination constant for HES 650/0.42/2.8. Regarding the global effects, HES 650/0.42/2.8 and HES 650/0.42/5.6 reach similar levels of absolute concentrations, hemodilutional effect, and alteration of blood coagulation, without differences between the groups.

The higher elimination constant found for HES 650/0.42/2.8 and the resulting smaller area under the curve, when compared with HES 650/0.42/5.6, confirm the fact that the C2/C6 ratio influences the HES degradation: The higher the C2/C6 ratio, the slower the HES is hydrolyzed due to a relative resistance toward α-amylolytic attack of the C2-linked side chains of the molecule.18,19Jung et al.  20found a prolongation of the elimination half time when increasing the C2/C6 ratio from 4.6 to 10.8, while keeping the same molecular weight of 200 kd and molar substitution of 0.5. We found a similar influence of the C2/C6 ratio on high-molecular-weight, low-substituted HES in the current study, resulting in a decreased blood exposure. This indicates that the influence of the C2/C6 ratio on HES pharmacokinetics is consistent over a considerable range of molecular weight. However, the range of C2/C6 ratio assessed in the current study is limited. To verify a general impact of the C2/C6 ratio on HES pharmacokinetics, a wider range of molar substitution including higher-substituted HES with molar substitutions of up to 0.6 or 0.7 and a wider range of C2/C6 ratios must be studied.

Treib et al.  18showed a significant difference regarding the in vivo  molecular weight after repeated infusion of HES 200/0.5/5.7 and HES 200/0.5/13.4 in patients. In that study, the HES with the higher C2/C6 ratio accumulated and had a higher in vivo  molecular weight, indicating a slower enzymatic breakdown. Contrarily to this, we did not observe higher concentrations 600 min after stop of infusion, possibly because of the lower substitution of the HES used.21Also, the lower level of C2/C6 ratio of the HES used in the current study may have contributed to a stronger cleavage by the α amylase. Moreover, we administered one single top-load dose without any redosing. In another crossover study with six patients who received either HES 200/0.5/10.8 or HES 200/0.5/5.8, the HES with the higher C2/C6 ratio had a higher plasma concentration and a higher in vivo  molecular weight after single-dose administration.14These results are partly confirmed by our study, as we also found significantly higher in vivo  molecular weights in the HES 650/0.42/5.6 than in the HES 650/0.42/2.8 group (figs. 2C and D).

In numerous studies, molecular weight is measured for the first time at the end of infusion. In these studies in vivo  molecular weight is at its maximum at the end of infusion and decreases subsequently.7,14,18,22We therefore expected during infusion an in vivo  molecular weight relatively close to the in vitro  molecular weight or at least between the in vitro  molecular weight and the in vivo  molecular weight at the end of infusion (30 min). However, we found that the in vivo  molecular weight increased during infusion to follow the expected decrease thereafter, i.e. , after the end of infusion at 30 min (figs. 2C and D). This phenomenon may suggest a metabolism mainly directed by α-amylase activity, which is increasingly saturated during HES infusion.

It remains to be elucidated whether the smaller area under the concentration–time curve for HES 650/0.42/2.8 compared with HES 650/0.42/5.6 is only due to higher intravascular metabolism and elimination rates or whether HES with a lower C2/C6 ratio may extravasate through the endothelial barrier23and subsequently be degraded by lysosomal α glycosidase24or other extravascular mechanisms.

Modifying the C2/C6 ratio did not affect the impact on blood coagulation (figs. 3 and 4). The major side effect of HES is the alteration of blood coagulation, which may be a serious limitation to its clinical use, especially in hemorrhagic patients or in those with coexisting coagulation disorders.25This was confirmed by our study, as several of the coagulation assays reflected the significant alteration of blood coagulation at the end of the HES infusion, which might become clinically relevant when used in humans. But, as mentioned above, molecular weight is not the only or the most important determinant of this side effect of HES. The isolated effect of molecular weight on blood coagulation has been assessed in a study using a top-load dose pig model, showing that high-molecular-weight, low-substituted HES (HES 900/0.4 and HES 500/0.4) influence plasma blood coagulation similarly as low-molecular, low-substituted-weight HES (HES 130/0.4).13The C2/C6 ratio was identical for these three solutions, as confirmed recently by the manufacturer. Also, the polydispersity index seems not to have a major influence on blood coagulation, because it does not represent molar substitution and or C2/C6 ratio, but only the distribution of molecular weight.

In combination with standard coagulation assays, TEG® is suitable to get a comprehensive assessment of coagulation changes during hemodilution (in vivo  as in vitro ) and in special clinical situations such as liver transplantation and cardiac surgery.9–11,13,26–33Nevertheless, TEG® is a global measure of the entire coagulation cascade and is therefore not suited to detect specific defects or deficits of the coagulation factors. However, TEG® is useful as a complementary measure to assess effects of volume expanders such as HES on clot formation and firmness.

The pig model is considered suitable for blood coagulation research.34Coagulation changes under various circumstances can be detected in the porcine model, such as in supraceliac aortic cross clamping, normovolemic hemodilution, and hemorrhagic shock.35–37There are differences as compared with the human coagulation system: elevated activity of FV, FVIII, FIX, FXI, and FXII resulting in a shortened aPTT indicating a slight hypercoagulability,34,38–40which was confirmed in our study. This hypercoagulability of the pig may have masked a part of the negative impact on blood coagulation of the HES solutions tested, and this may be a limitation of our study. However, we were capable of detecting a highly significant compromise of blood coagulation due to the infusion of HES. The absence of differences between the groups thus is not due to a lack of sensitive measures but due to the fact that there is no relevant difference between the HES solutions tested. In addition, McLoughlin et al.  36have shown that the response of prothrombin time and aPTT to profound hemodilution is similar in humans and in pigs. Therefore, we may assume that the porcine coagulation reacts relatively similar to hemodilution with HES solutions as compared with the human coagulation system.

Hemodilutional effects, as measured by the indirect markers hemoglobin and albumin, were not different between the study group and the control group (fig. 5). This is not a surprising finding, despite the fact that the area under the time–concentration curves and the elimination constants showed significant differences between the solutions. In fact, water binding capacity is related to the number of the osmotically active particles.3,5The number and size of HES molecules in vivo  is changing constantly over time from the beginning of infusion.3,4In our study, we confirmed these findings: HES 650/0.42/2.8 has in vivo  significantly lower molecular weights than HES 650/0.42/5.6 because of the facilitated cleavage by α amylase, which results in a decreased area under the curve. Because of the equal number of osmotically active particles, the same volume-expanding effect is mediated by the two HESs (figs. 2B and 5). Also, the molar substitution and C2/C6 ratio of the HES solution change constantly in the organism and could therefore influence the volume-expanding properties.4 

Our findings regarding hemodilution are limited by the fact that the study was not specifically designed to directly detect plasma volume–expanding effects. To determine exactly volume changes in the organism, more sophisticated methods must be applied, such as double-label measurements of blood volume.23Beyond this, HES 650/0.42/2.8 must prove its volume-expanding effect in clinical more relevant models, such as redosing studies7,18,41or animal models of controlled hemorrhage, which have been shown to be adequate for the comparison of the effectiveness of different resuscitation fluids.42–46 

We conclude that lowering of the C2/C6 ratio of high-molecular-weight, low-substituted HES 650/0.42/5.6 to HES 650/0.42/2.8 results in a faster cleavage and elimination from the intravascular space, without any evidence for a reduction of the volume-expanding effect. Because the blood coagulation compromising effect of the two solutions was similar, the specific impact of the C2/C6 ratio of high-molecular-weight, low-substituted starches on blood coagulation seems to be relatively small. More research remains to be conducted with HES 650/0.42/2.8 to evaluate its clinical benefit.