Artificial Cells, Blood Substitutes, and Biotechnology, 34: 567580, 2006 Copyright Q Informa Healthcare ISSN: 1073-1199 print/1532-4184

online DOI: 10.1080/10731190600973824

Perfluorocarbon-based Oxygen Delivery

Jean G. Riess
MRI Institute, University of California at San Diego and Alliance Pharmaceutical Corp., San Diego, CA, USA

Abstract: The basic properties of perfluorocarbons (PFCs) and PFC emulsions relevant to their use as oxygen delivery systems are briefly reviewed. The key issues related to the selection of an appropriate, readily excretable PFC and the engineering of a stable injectable PFC emulsion are discussed. OxygentTM, a terminally heat-sterilized, injectable 60% w=v PFC emulsion made primarily of F-octyl bromide and a few percent of F-decyl bromide, with egg phospholipids as an emulsifier, has been developed. Its efficacy in avoiding and reducing red cell transfusion during surgery has been established during a Phase III clinical evaluation. Another Phase III clinical trial in cardiopulmonary bypass surgery, with a protocol that included both augmented-acute normovolemic hemodilution and intraoperative autologous donation, has, however, been interrupted following the observation of adverse events. Data analysis assigned these events to an inappropriate study protocol. A search for possible interactions between Oxygent and fluids present during cardiopulmonary bypass surgery detected no effect of the emulsion on hemostasis, hemolysis and blood rheology.

INTRODUCTION This short review, which is based on a plenary lecture given at the Xth Int. Symp. on Blood Substitutes at Providence, comprises brief reminders about the properties of perfluorocarbons (PFCs) and PFC emulsions relevant to their use for oxygen delivery; a discussion of the key issues related to the development of an effective injectable product; the clinical evaluation status of Oxygent; and a summary of extensive studies of the effects of Oxygent on hemostasis and viscosity of bloodcontaining fluids present in the patient during a cardiopulmonary bypass trial.
Address correspondence to Jean G. Riess, Alliance Pharmaceutical Corp., 4660 La Jolla Village Drive #825, San Diego, CA 92103, USA. E-mail: jriess@allp.com 567

568

J. G. Riess

BASIC REMINDERS PFCs offer unique combinations of high O2 and CO2 dissolving capacities, low water and lipid solubilities, and exceptional chemical and biological inertness [13]. Gas solubility reflects the very low intermolecular interactions (fluorines low polarizability translates into low van der Waals forces) within a PFC liquid. Figure 1 schematically compares the O2 dissolution and binding mechanisms that underlie O2 transport by fluorocarbons and hemoglobin (Hb). In the case of Hb, a strong, localized chemical coordination bond is established between the dioxygen molecule and the iron atom of a heme. In the case of PFCs, there is physical dissolution of O2, characterized by loose, non-directional van der Waals interactions among like materials. Indeed, PFCs and gases both have very low cohesive energy densities, as expressed by very close Hildebrandt coefficients. The difference in nature of these interactions is clearly reflected by the differences in profiles of the O2 uptake curves as a function of pO2, i.e. sigmoidal for Hb vs. linear for PFC emulsions. With PFCs, there is no saturation and no possibility for chemical binding of, and interference with, NO, CO or other reagents. PFC-dissolved O2 is immediately available to tissues; it is also characterized by high extraction ratios. Furthermore, O2 dissolution and release to tissues is preserved when temperature decreases.

Figure 1. Oxygen transport by perfluorochemicals and hemoglobin: in the case of Hb, O2 is bound to the iron atom of a heme through a strong, localized chemical bond. In PFCs, there exist only loose, non-directional van der Waals interactions; PFCs and gases are alike; both present very low cohesive energy densities, which facilitates mutual solubilization.

Perfluorocarbon-based Oxygen Delivery

569

PFCs outstanding stability and inertness relate to strong CF bonds and to the dense, protective and repellent electron sheath that coats F-chains (F- stands for perfluoro). Low water and lipid solubilities reflect the hydrophobic and lipophobic characters of PFCs. PFCs tend therefore to segregate (i.e. keep to themselves) and do not meddle with other material, which also contributes to their biological inertness. These hydrophobic and lipophobic effects promote phase separation, selfassembly and ordering among fluorinated components. The biological inertness of PFCs is illustrated by the fact that ingestion of F-octyl bromide in liter-size doses has been approved by the FDA for use as a contrast agent for bowel delineation by MRI. Likewise, filling the lungs of a patient with a liquid PFC can be done without serious untoward effects. Further PFC-specific properties potentially useful in biomedical applications include extremely low surface tensions, high density, fluidity and spreading coefficients, absence of protons, and a concentration of 19 F nuclei that provides a valuable NMR probe. F-materials are unique components for the engineering of a large variety of stable colloidal systems with highly fluorinated two or threedimensional phases [2,4]. These colloids consist typically of dispersions of spherical micro and nano-domains with internal phases separated from a distinct external phase by some shell or membrane. PFCs and PFC-based colloids are being investigated for a variety of biomedical applications [5,6]. Examples of such colloids (Fig. 2) include the commercial, injectable PFC-stabilized, micron-size gas bubbles that serve as contrast agents for ultrasound imaging [7,8] and the sub-micron size liquid PFC droplets that are being developed for O2 delivery [9,10].

DEVELOPMENT OF A PFC EMULSION-BASED OXYGEN CARRIEROXYGENT Development of a PFC emulsion-based O2-carrier involves selection of an appropriate PFC and the engineering of a stable injectable emulsion [9]. PFCs for O2 delivery require relatively low vapor pressure, rapid excretion, and the capacity of forming stable emulsions. They also need to be easy to synthesize on a large scale in highly pure form at a reasonable price, which implies the availability of a large-tonnage industrial precursor and an appropriate synthetic process. Combining high emulsion stability and rapid excretion is essential and not common among PFCs. Slightly lipophilic PFCs, such as F-octyl bromide (perflubron), were shown to fulfill these conditions. F-octyl bromide has, for example, a critical solution temperature in hexane (a measure of lipophilicity) that is ca. 40C lower than that of F-decalin, in spite of its higher molecular

570

J. G. Riess

Figure 2. Two examples of injectable colloids with a fluorous internal phase: PFC-stabilized gas bubbles used as contrast agents for diagnostic ultrasound imaging and PFC emulsion droplets developed for oxygen delivery to tissues.

weight (499 vs. 462; MW is the principal determinant of excretion rate), and an 8 times larger solubility in olive oil. Excretion half-lifes for similar injected doses are 4 days for the former PFC, as compared to 7 days for the latter. Manufacture for parenteral administration requires that the PFC be pure, well defined, reproducible, derived from an industrially available starting material, and scalable. These requirements are more easily met with a telomerization process (e.g. telomerization of tetrafluoroethene), rather than by a fluorine=hydrogen substitution process, such as electrochemical fluorination, CoF3 treatment or direct F2 fluorination. F-octyl bromide can be manufactured in large tonnages and better than 99.9% purity by direct bromination of F-octyl iodide in one step. F-octyl iodide is obtained by telomerization of tetrafluoroethylene, the precursor of Teflon, and is itself the precursor of many industrial surfactants. Intravascular administration of PFCs requires the engineering of a stable, terminally heat-sterilized, ready-for-use submicronic emulsion. This emulsion needs to be stable in standard, non-frozen conditions for at least two years. It needs to have small particles and a narrow particle size distribution. Small droplets, typically in the 0.10.2 mm range, ensure prolonged intravascular persistence, facilitated O2 diffusion and minimal side effects. Narrow size distributions and the absence of a tail of larger droplets result in slower molecular diffusion, and hence increased emulsion stability, as well as reduced macrophage activation.

Perfluorocarbon-based Oxygen Delivery

571

Dispersion of a highly hydrophobic PFC in water (interfacial tension ci 5060 mN m1) requires an effective surfactant and high energy input. Egg yolk phospholipids proved effective in reducing ci. Phospholipids are commonly used in pharmaceuticals, including in the closely related lipid emulsions used for parenteral nutrition. High shear is provided by a high pressure Gaulin-type homogenizer. Terminal thermal sterilization and large-scale production capability with batch-to-batch consistency in cGMP conditions are mandatory. Large-scale parenteral emulsion manufacturing technology is available and the know-how specific to PFCs has been developed. Emulsion manufacture being an additive process, its yield is nearly 100%. Droplet growth over time in PFC emulsions is driven by molecular diffusion (also known as Ostwald ripening and isothermal distillation) rather than coalescence. In this mechanism, individual molecules leave the smallest droplets, in which the chemical potential is higher due to higher curvature, and join larger droplets. Molecular diffusion is characterized by a linear increase of droplet volume over time and by a time-invariant size distribution function. The rate of increase of the droplets volume over time is a linear function of, among others, the interfacial tension and the solubility and diffusibility of the dispersed PFC in the aqueous phase. Emulsion stability increases with MW, as water solubility decreases and, consequently, Ostwald ripening of emulsion droplets becomes slower. On the other hand, the PFC excretion rate decreases with increasing MW, as lipid solubility, which plays a role in the in vivo transport of PFCs, decreases. Molecular diffusion can be effectively slowed down by incorporation of small amounts of a less water-soluble, higher MW secondary PFC, at the expense, however, of slower excretion. Use of a slightly lipophilic secondary PFC prevents excessive organ retention. In the case of F-octyl bromide emulsions, the rate of droplet volume growth could be reduced by a factor of ca. 5 by simply adding a few percent of F-decyl bromide, a higher homologue of F-octyl bromide. In vivo PFC distribution and elimination are primarily characterized by two parameters (Fig. 3), namely PFC half-life in the circulation (t1=2) and PFC half-life in the RES organs (T1=2). The PFC is not metabolized and its excretion occurs through the lungs with the expired air. For a given dose, t1=2 (typically in the 824 h range in humans) depends primarily on the characteristics of the droplets, including size and nature of its coating, while T1=2 (typically in days and weeks) depends on PFC characteristics, i.e. MW and lipophilic character. Being foreign particles, PFC emulsion droplets are eliminated by the RES. Inadequate size and surface characteristics can cause macrophage activation that triggers flu-like symptoms. This elimination mechanism is also responsible for

572

J. G. Riess

Figure 3. In vivo PFC distribution and elimination. Circulatory half-life (t1=2) of a PFC emulsion depends essentially on droplet characteristics, while retention in the RES organs (T1=2) depends primarily on PFC characteristics.

the relatively short intravascular persistence of the present emulsions, which limits their use to perisurgical-type applications. Extensive development efforts led to Oxygent AF0144 (Alliance Pharmaceutical Corp.), a 60% w=v concentrated PFC emulsion, primarily perflubron, with a few percent of F-decyl bromide added as an Ostwald ripening repressor, emulsified with egg yolk phospholipids [9]. The emulsion is heat sterilized in a rotary sterilizer. It is pyrogen and endotoxin negative. Average droplet size is 0.16 mm and viscosity around 4 cP. Oxygen solubility is ca. 17 vol.%. Molality is adjusted to 304 mOsm. A unit dose of the product is provided as 110 mL of emulsion containing 65 g of PFC in a glass bottle. It is stable for 2 years at 510C and ready for use. Clinical trials in general surgery have established that such a unit dose is about equivalent to one unit of red blood cells [11]. Preclinical and clinical evaluation of Oxygent established its ability to deliver oxygen to tissues [12]. A Phase III trial in general surgery, with an augmented acute normovolemic hemodilution (A-ANH) protocol, demonstrated efficacy through statistically significant transfusion avoidance and transfusion reduction [13]. A subsequent trial in coronary artery bypass grafting patients undergoing cardiopulmonary bypass (CPB) surgery with an A-ANH plus intraoperative autologous donation (IAD) protocol uncovered, however, untoward bleeding and neurological events that caused Alliance to suspend the trial. Thorough analysis of the clinical safety data led to the conclusion that the study conductand not the emulsionwas responsible for the observed adverse events. It was statistically demonstrated that the imbalance in neurological and bleeding events was caused by excessive hemodilution in the patients receiving Oxygent. There was also an association between the occurrence of these events and the investigational sites experience with the dilution procedures [12,14]. The study protocol used

Perfluorocarbon-based Oxygen Delivery

573

Figure 4. Schematic representation of the study protocol used during a Phase III clinical evaluation of Oxygent in cardiopulmonary bypass surgery. The Oxygent test group underwent both acute normovolemic hemodilution (ANH) and autologous intraoperative donation (IAD). The time points when the IBB and IAD fluids investigated would be present in the patient are represented by the horizontal arrows.

is schematically represented in Figure 4. According to this protocol, both the Oxygent test group and the control group underwent ANH. The test group then received two thirds of the intended clinical dose of 2.7 g=kg body weight of Oxygent. After having been put on bypass, these patients (but not the patients in the control group) underwent an additional withdrawal of approximately one liter of their blood (now containing some of the prime solution used for filling the bypass circuit) in an IAD procedure. They then received the remainder of the dose of Oxygent. Inadequate management of blood pressure during the rapid autologous blood harvesting procedure potentially resulted in decreased perfusion to the brain in some patients at greater risk for complications [12,14]. New trials are being initiated in collaboration with Double Crane Pharm. Co. (Beijing, China). Other trials are being started in Europe, the indication being to decrease the incidence of post-operative organ failure. HEMOSTASIS, HEMOLYSIS AND RHEOLOGY STUDIES An extensive search has been undertaken for possible chemical or physical interactions between Oxygent and the blood-containing fluids present in the patient during CPB surgery (Fig. 4). More specifically, we looked for any perturbation of hemostasis or increase in viscosity that could be related to the observed bleeding events. A reduction in speed of clot

574

J. G. Riess

formation, reduction of clot strength or increased rate of lysis would be undesirable and might lead to postoperative bleeding. The opposite effects could lead to a greater tendency to clotting and to embolic neurological phenomena. An increase in viscosity, particularly at low shear rates, might be undesirable because it could impede blood flow in the microcirculation. The two test fluids investigated, representative of those encountered during the trials, consisted of: a) A fluid that mimicked the blood-containing fluid present in the patient immediately before bypass (IBB fluid); this fluid comprised human blood (76%), colloid (20%), heparin (0.043 mg=mL) and Oxygent (4% 1.8 g=kg dose). b) A fluid that mimicked the blood-containing fluid that is drawn from the patient after intraoperative autologous donation (IAD) has been performed. It comprised blood (45%), colloid (12%), Duke prime (40%), heparin (0.025 mg=mL) and Oxygent (2.5%) (Duke Prime: 200 mL colloid; 20 mL mannitol [20%], 20 mEq HNaCO3, 2 mL heparin [1000 units=mL], 780 mL Plasma-Lyte A; colloid was 6% hetastarch in 0.9% NaCl solution). The control fluids consisted of the same fluids, but with a phosphate buffer system (PBS) instead of Oxygent, or a soybean oil emulsion (SOE) in which the PFC was substituted by soybean oil, and of whole blood (WB). Measurements were made both on fresh mixtures and on mixtures incubated for 4 h with Oxygent, PBS or SOE.The incubated mixtures also mimicked the emulsion-containing IAD blood waiting for reinfusion. Fresh human blood was drawn into heparin or citrate, the latter being representative of banked blood. Hemostasis was investigated using thromboelastography (TEG; Fig. 5). TEG assays the dynamic rheology of fibrin clots. It measures the evolution of the shear modulus over time during the clotting process and subsequent lytic events [15]. The following parameters were measured: reaction time, i.e. time until initial fibrin formation; rate of clot formation; clot strength; and clot lysis after 30 min. Coagulation was stimulated by protamine sulfate in the case of heparinized blood, and by protamine sulfate and CaCl2 in the case of citrated blood; celite was added, which activates Factor XII [16]. A typical example of the data collected is shown in Figure 6, which represents reaction time, rate of clot formation, clot strength and 30 min clot lysis of Oxygent-containing IBB and IAD fluids, as compared to the same fluids, but with PBS or SOE instead of Oxygent. The samples were again investigated after a 4h-incubation period. The values obtained

Perfluorocarbon-based Oxygen Delivery

575

Figure 5. Thromboelastography provides a set of four parameters that characterize the formation and lysis of fibrin clots: reaction time (R, min); rate of clot formation (a, d); clot strength (max. amplitude MA, mm); clot lysis 30 min after MA (% of MA).

for WB and WB diluted with an hydroxyethylstarch solution are provided for reference. The results from the IBB=heparinized mixtures showed no significant differences, for any of the four parameters measured, between the Oxygent test mixture and the control mixtures. In particular, there were no differences between the IBB=Oxygent and the IBB=PBS mixtures, whether fresh or after 4 h incubation. The 3=1 blood=colloid mixture took slightly, but not significantly, longer to begin clotting. The rate of clot formation was, as expected, higher in the WB reference compared to that of the other mixtures, which were all equally 75 vol% diluted blood mixtures. No differences in clot strength were observed between test and control groups. Clot strength was, again as expected, greater for the WB reference than for any of the other groups. Likewise, the heparinized mixtures had all similar clot lysis values, irrespective of the presence or not of Oxygent. The IAD fluid mixtures being more red cell-depleted (45% blood) coagulated significantly more slowly than WB; the 4h-incubated IAD mixtures coagulated somewhat faster than the fresh fluid. Most importantly, there were no significant differences between the Oxygent-containing group and the PBS or SOE groups, whether fresh or incubated. Likewise, there were no differences in rate of clot formation, clot strength or clot lysis

576

J. G. Riess

Figure 6. Effect of Oxygent on coagulation a) of the blood-containing fluid circulating in patients immediately before bypass (IBB fluid, see Fig. 4), and b) of the blood-containing fluid present in patients after IAD has been performed (IAD fluid, see Fig. 4). The human blood used was heparinized. WB whole blood; PBS phosphate buffer system; SOE soy bean emulsion (courtesy Alliance Pharmaceutical Corp.).

Perfluorocarbon-based Oxygen Delivery

577

between the PFC emulsion group and the two control groups, whether in fresh fluids or after 4 h incubation. Oxygent did not interfere either with coagulation of citrated IBB or IAD fluids. In conclusion, the hemostasis study of Oxygent AF0144containing IAD and IBB blood fluids, heparinized or citrated, fresh or after 4 h standing, found no effect of Oxygent on hemostasis different than those of PBS. Hemolysis of heparinized porcine blood in IBB and IAD fluid mixtures was investigated under constant shear stress for up to 2 h (Fig. 7). Measurements of the plasma concentration of free Hb revealed no significant differences between the Oxygent- and PBS-containing mixtures. All

Figure 7. Swine red blood cell hemolysis in a) IBB and b) IAD mixtures under constant shear stress (0.5 Pa, 37C, n 3). The mixtures contained either Oxygent or PBS (courtesy Alliance Pharmaceutical Corp.).

578

J. G. Riess

the values leveled off rapidly and remained below those typically reported during CPB surgery. A search for possible effects of the PFC emulsion on the rheology of the circulating fluids was also undertaken. The rheology studies included constant shear (steady state) viscosimetry, shear-dependent viscosimetry and yield stress measurements [16]. A control stress rheometer with a double-gap sample cell was used and the measurements were made on porcine blood on IAD or IBB mixtures containing either AF0144 or one of two control fluids, i.e. PBS or SOE. Measurements were also performed on whole blood and on a 3:1 blood:colloid mixture as references. Figure 8 exemplifies constant shear stress viscosity data. Viscosity was measured as a function of shear rate under steady state conditions. Figure 8a shows that all the IBB mixtures investigated had lower viscosities than WB. The lowest values were found for the samples incubated with Oxygent for 4 h. The viscosities of the IAD mixtures were lower than those of the IBB mixtures, as expected due to lower blood fraction (45 vs. 76 vol.%). The differences between the various samples were more pronounced at low shear rates (Fig. 8b). The lowest viscosity was again found for the samples incubated with Oxygent for 4 h. All the mixtures were considerably more fluid than WB. Yield stress of IBB and IAD fluid mixtures were identical whether they contained Oxygent or PBS or none of them. The stress viscosimetry measurements indicated no significant differences between the IBB mixtures with and without Oxygent or PBS, fresh or after 4 h, and WB. For the IAD mixtures, the viscosity values were essentially identical for the IAD=Oxygent and IAD=PBS and IAD without Oxygent or PBS samples, fresh or after 4 h. As expected, all the values were significantly lower than for WB, especially at low shear stress. No differences in yield stress (minimum shear stress needed to induce flow) measurements were found between IBB alone, IBB=Oxygent fresh or after 4 h incubation, and IBB=PBS. Likewise, the IAD mixtures with and without Oxygent had the same yield stress. Comparing the IBB and IAD data, it appears clearly that the main contributor to yield stress is, not unexpectedly, the blood fraction present in the mixture. The higher the latter, the higher the shear stress that is needed to induce flow. Whether the blood has been diluted by the fluorocarbon emulsion or by a simple buffer solution is inconsequential. All the rheological measurement modes investigated have established that Oxygent had no negative influence on the viscosity of the IBB and IAD blood fluids, different from that of PBS. Viscosity was, in all cases, significantly reduced as compared to WB (reflecting the reduced WB fraction in the IBB and IAD fluids), irrespective of the presence or absence of the PFC emulsion. Actually, the Oxygent mixtures provided the best fluidity characteristics.

Perfluorocarbon-based Oxygen Delivery

579

Figure 8. Shear rate-dependent viscosity of a) IBB fluid mixtures, and b) IAD fluid mixtures (courtesy Alliance Pharmaceutical Corp.).

The rates of passive filtration through 5 and 10 mm filters under gravity of IBB and IAD fluid mixtures and control fluids were also determined. These conditions mimic the flow conditions in capillaries. The IBB=Oxygent and IBB=PBS samples were found to flow through the filters at similar rates, indicating that the emulsion did not impede the flow any more than the buffer solution and that the aggregation forces that

580

J. G. Riess

exist between emulsion droplets are lower than those at work during filtration, even under minimal gravity flow. The IAD=Oxygent sample actually traversed the 5 mm filter more rapidly than the PBS control, suggesting that Oxygent facilitated, rather than hindered, the flow. Microscopic observations indicated that the emulsion droplets did not tend to aggregate with red blood cells. IN SUMMARY No chemical=physical interaction between Oxygent AF0144 and the IAD or IBB fluids was detected. No perturbation of normal hemostatic or viscosity behavior was found that could be relevant to the neurological and bleeding events observed during the Phase III CPB trial. There was no reduction in speed of clot formation, reduction of clot strength or increased rate of lysis that could result in postoperative bleeding. Neither did the presence of Oxygent induce any increased tendency for coagulation that could have an effect on the occurrence of thrombotic strokes. The PFC emulsion did not increase viscosity (it tended actually to reduce it) and hence is unlikely to impede microvascular perfusion. REFERENCES
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. Krafft, M.P., Riess, J.G. (1998). Biochimie. 80: 489514. Riess, J.G. (2002). Tetrahedron 58: 41134131. Riess, J.G. (2005). Artif. Cells Blood Subst. Immob. Biotech. 33: 4763. Krafft, M.P. (Ed.) (2003). Curr. Opin. Colloid Interf. Sci. 8: 213. Krafft, M.P. (2001). Adv. Drug Deliv. Rev. 47: 209228. Riess, J.G. (2004). in Handbook of Fluorous Chemistry, Gladysz, J.A., Curran, D.P., Horv ath, I., Eds., Wiley-VCH: Weinheim, pp. 521573. Schutt, E.S., Klein, D.H., Mattrey, R.M., Riess, J.G. (2003). Angew. Chem. Int. Ed. 42: 32183235. Riess, J.G. (2003). Curr. Opin. Colloid Interf. Sci. 8: 259266. Riess, J.G. (2001). Chem. Rev. 101: 27972920. Krafft, M.P., Chittofrati, A., Riess, J.G. (2003). Curr. Opin. Colloid Interf. Sci. 8: 251258. Faithfull, N.S. (2003). Adv. Exp. Med. Biol. 530: 271285. Keipert, P.E. (2005). in Blood Substitutes, Winslow, R.M., Ed., Elsevier: Amsterdam., chap. 28, pp. 312323. Spahn, D.R., Waschke, K.F., Standl, T., Motsch, J., Van Huynegem, L., Welte, M., Gombotz, H., Coriat, P., Verkh, L., Faithfull, S., Keipert, P. (2000). Anesthesiology 97: 13381349. Faithfull, N.S. (2005). Int. Symp. Blood Substitutes, Providence, RI. Luddington, R.J. (2005). Clin. Lab. Haematol. 27: 8190. Le, T., Obraztsov, V.V., Seang, Y., Smith, D., Riess, J.G. Unpublished.

14. 15. 16.