Use of Perflubron Emulsion to Decrease Allogeneic Blood Transfusion in High-blood-loss Non-Cardiac Surgery: Results of a European Phase 3 Study

The study was performed at 34 centers in 8 European countries from November 1998, to June 2000. All local ethics committees approved the protocol. After we obtained written informed consent, 492 patients undergoing major noncardiac surgical procedures with expected blood loss between 20 and 70 ml/kg were enrolled. Additional inclusion criteria included preoperative hemoglobin concentration between 12 and 15 g/dl, age between 18 and 80 yr, weight between 50 and 125 kg, American Society of Anesthesiologists physical status I–III, and estimated blood volume sufficient to allow removal of at least 2 units of autologous blood during preoperative ANH.

Exclusion criteria were pregnancy or lactation; refusal of allogeneic erythrocyte transfusion; history of myocardial infarction within 6 months; unstable angina or coronary artery disease, placing the subject at risk of myocardial ischemia at the hemoglobin concentration defined by the protocol (8.0 ± 0.5 g/dl); severe chronic obstructive pulmonary disease or other pulmonary condition placing the subject at risk due to low hemoglobin concentration and inability to substantially increase their arterial oxygen partial pressure with a fraction of inspired oxygen (Fio²) of 1.0; carotid artery disease or history of transient ischemic attacks or amaurosis fugax; systemic infection, clinical signs of sepsis, leukocytosis, or fever greater than 38.5°C; trauma within 72 h of surgery; history of bleeding disorder; preoperative platelet count less than 150,000/μl; significant hepatic disease, defined as aspartate aminotranferase or alanine aminotranferase greater than twice the upper limit of normal; significant renal disease, defined as a creatinine concentration greater than 180 μm/l; use of cell salvage; pharmacologically induced hypotension; participation in other studies involving an investigational drug or device within 30 days or within 12 months when an artificial oxygen carrier was involved; and history of hypersensitivity to egg yolk or any constituent of PFC.

This was a prospective, multicenter, single-blind, randomized, controlled, parallel-group, phase 3 study. PFC (Oxygent™; Alliance Pharmaceutical Corp., San Diego, CA) is a 60% wt/vol perfluorochemical emulsion based on perflubron emulsified with 3.6% wt/vol egg lecithin in phosphate-buffered saline, and was provided by the manufacturer, who also sponsored the study. After providing written informed consent, each patient was randomized within 48 h before surgery to either standard of care (control) or PFC treatment in conjunction with ANH. All patients had radial artery and central venous catheters in addition to standard anesthesia monitoring. Neither anesthetic drugs nor infusions (crystalloids and colloids) were specified by protocol. Hemoglobin concentration (HemoCue; AB Leo Diagnostics, Helsinborg, Sweden) was measured after each unit of blood removed during ANH (PFC group) and intraoperatively at least every 30 min.

Patients were maintained at an Fio²of 0.4 and transfused with erythrocytes from preoperative autologous donation (PAD), if available, or allogeneic erythrocytes for each intraoperative transfusion trigger, i.e. , either a hemoglobin concentration less than 8.0 ± 0.5 g/dl or any one of the protocol-defined intraoperative physiologic triggers not easily reversed by fluid adminiatration or adjustment of anesthesia. These included tachycardia (heart rate ≥ 100 beats/min or ≥ 135% of postanesthesia induction value), hypotension (mean arterial pressure ≤ 60 mmHg or ≤ 65% of postinduction value), mixed venous oxygen partial pressure 38 mmHg or less (if a pulmonary artery catheter was used), or ST-segment depression (> 0.1 mV) or elevation (> 0.2 mV).

Prior to surgical incision, patients underwent ANH to a hemoglobin concentration of 8.0 ± 0.5 g/dl at an Fio²of 1.0. Investigators were free in their choice of standard crystalloids and colloids to replace the blood withdrawn during ANH according to recommended guidelines for volume replacement. Clinical signs of normovolemia (stable heart rate and blood pressure) further guided amounts of replacement fluids administered during ANH; central venous pressure was not monitored systematically during ANH but was available later to help assess volume status. At skin incision, a 1.8-g/kg dose (3 ml/kg) of PFC was given. This represents approximately 0.94 ml/kg of the PFC active ingredient, i.e. , nonemulsified perfluorochemical. When hemoglobin reached 6.5 ± 0.5 g/dl during surgery, a second 0.9-g/kg (1.5 ml/kg) dose of PFC was administered (∼0.47 ml/kg of neat perfluorochemical). When hemoglobin concentration was less than 5.5 ± 0.5 g/dl or when the physiologic triggers (described in the previous paragraph for the control group) were met, patients were transfused with ANH or PAD units, if available, before receiving allogeneic erythrocytes. All units of autologous blood collected during ANH were retransfused after surgery.

In both groups, a hemoglobin concentration of 8.5 ± 0.5 g/dl or greater was targeted at the end of surgery, maintained through postoperative day 3, and was required by protocol as the minimum concentration at hospital discharge. The ranges in hemoglobin transfusion triggers (± 0.5 g/dl) were not applicable for individual patients but were provided to allow centers to adjust transfusion decisions (equally in both groups) according to local standards. Investigators were required to measure hemoglobin concentrations postoperatively to ensure that protocol-mandated transfusion requirements were followed, but this person was not blinded to the group assignment of the patient. No specific criteria for the perioperative administration of fresh frozen plasma, cryoprecipitate, or platelet transfusions were defined in the protocol.

An intraoperative hemoglobin trigger difference of 2.5 g/dl between control and PFC-treated groups was based on pharmacodynamic and pharmacokinetic data from previous phase 2 studies. These data included physiologic efficacy of perflubron emulsion in orthopedic and urologic patients receiving a 1.8-g/kg dose of PFC (details provided in  Appendix 2). Physiologic modeling of a 2.7-g/kg dose predicted that a hemoglobin equivalency of 2.5 g/dl or greater after blood loss would decrease hemoglobin to 5.5 g/dl. As a result, PFC-treated patients with a hemoglobin concentration of 5.5 g/dl have similar oxygen dynamics as control patients with a hemoglobin concentration of 8.0 g/dl.

Blood samples were collected in both groups for laboratory analyses at several time points: after anesthesia induction but prior to ANH, at arrival in the recovery room or intensive care unit, and on postoperative day (POD) 1, 2, 3, 7, and 21 or day of discharge (DD), whichever came first. Patients were also contacted 3 months after surgery to obtain additional follow-up information regarding postoperative adverse events (AEs). AEs and serious AEs were reported according to standard regulatory guidelines for current Good Clinical Practice. No predefined definitions of clinically relevant AEs were provided to investigators. An independent data safety monitoring board was used to periodically review laboratory and safety data during the conduct of this study.

The sample size for this study was based on the primary efficacy endpoint, with a type 1 error (α) = 0.049 (adjusted from 0.05 due to the interim analysis), and power (β) = 90%. The minimum clinically meaningful difference in mean reduction in the number of erythrocyte units transfused is δ= 1.0. The estimate of the pooled SD for a comparison of the mean difference in the number of erythrocyte units transfused is S^p= 3.0; this estimate was obtained from phase 2 clinical studies. A two-sample t  test for sample size estimation yielded n = 2 (Zα^/2+ Zβ)²S^p2/δ²= 2 (1.97 + 1.28)²· 9/1.0, about 190 per group. Assuming a 5% dropout rate increased this to 200 per group, and allowing for 20% of subjects to have a blood loss less than 20 ml/kg increased the required sample size to about 240 per group.

The primary efficacy endpoint for this protocol was the number of allogeneic or PAD units transfused during the acute study period (24 h following skin incision). Secondary endpoints included the percentage of subjects avoiding allogeneic erythrocyte transfusions during the acute study period, the percentage of subjects avoiding allogeneic erythrocyte and PAD unit transfusions during the acute study period, and the elapsed time from immediate arrival in the recovery room to DD or POD 21. In addition, transfusion outcome (number of allogeneic or PAD units transfused and percentage of patients avoiding any allogeneic or PAD transfusion) on PODs 1, 3, and 7 and on DD (or POD 21, whichever came first) was prospectively recorded and analyzed. The primary efficacy population was defined as the intent-to-treat population, which included all randomized subjects. The secondary efficacy population prospectively defined in the protocol was all randomized subjects with estimated blood loss of 20 ml/kg or greater. The safety population included all treated subjects who underwent surgery and, if randomized to receive PFC, who also received at least the first dose of PFC.

Treatment groups were compared for the number of erythrocyte or PAD units transfused using an analysis of covariance, with treatment group, site, and screening hemoglobin in the model. A rank transformation of the number of units transfused was used to be able to account for outliers. To estimate the mean and mean difference effects, a log transformation was used. A logistic regression was used to compare treatment groups for the number of subjects avoiding erythrocyte and PAD transfusions. Treatment group, site, and type of surgery were in the model. For demographic and safety data, treatment groups were compared using the Fisher exact test for categoric data and t  tests for continuous data. All data are presented as means ± SD unless otherwise indicated.

Patient demographics were similar in both groups at screening and baseline (table 1), as were types of surgeries (table 2). A total of 24 patients (14 PFC and 10 controls) were withdrawn before treatment or surgery, either at the subject's request or because clinical condition did not warrant surgery at the time (these subjects are all included in the efficacy analysis for the intent-to-treat group but are excluded from the safety analysis). During preoperative ANH, 1,618 ± 558 ml (range, 450–3,374 ml) of blood was withdrawn and replaced by 1,312 ± 680 ml (range, 100–3,500 ml) colloid and 2,418 ± 1627 ml (range, 100–10,000 ml) crystalloid infusions. The mean post-ANH hemoglobin concentration achieved was 8.1 ± 0.5 g/dl. All 241 PFC-treated patients received the 1.8-g/kg dose of PFC, but only 177 subjects lost enough blood to require the 0.9-g/kg dose. Hemoglobin concentrations, per protocol, were lower intraoperatively in the PFC group (P < 0.001) but returned to similar values postoperatively (fig. 1).

In the intent-to-treat population (n = 492), the PFC group (n = 241) required fewer transfusions than controls (n = 251) despite a higher estimated intraoperative blood loss (2.7 ± 2.7 vs.  2.3 ± 2.0 l, respectively;P < 0.05). The primary endpoint (reduction in number of allogeneic–PAD units transfused at 24 h) was achieved: the PFC group received 1.5 ± 4.8 versus  2.1 ± 3.9 units (median 0 vs.  1 unit) in controls (P = 0.013), representing a reduction of 26% (table 3). After POD 3, although differences were still present, they were no longer statistically significant. However, in the protocol-defined target population (blood loss ≥ 20 ml/kg; n = 330, or 67% of all randomized subjects), the PFC group had a larger reduction in allogeneic and PAD units transfused (mean, 2.0 ± 4.0 vs.  3.3 ± 3.0; median, 1 vs.  3 units;P < 0.001) on POD 1 (table 3), and this difference remained significantly different from controls through discharge (mean, 3.4 ± 2.9 vs.  4.9 ± 2.4; median, 2 vs.  4 units at POD 21 or DD;P < 0.001). In total, for the intent-to-treat population through POD 21 or DD, PFC-treated subjects required 696 units versus  846 units in the control group.

Regarding complete avoidance in the intent-to-treat population, approximately 21% more patients in the PFC group avoided allogeneic and PAD transfusions compared with controls (P < 0.05) during the acute study period (24 h). Later, the difference was not significant any more (fig. 2). However, in the protocol-defined target population, a significantly (P < 0.05) greater percentage (almost twice as many) of patients avoided transfusion at all time points from POD 1 through POD 21 or DD (fig. 2). A post hoc  analysis to identify patients who benefited from PFC treatment indicated that when estimated surgical blood loss was 10 ml/kg or greater (n = 424; 86% of all randomized subjects), transfusion was significantly reduced in PFC-treated patients versus  controls at all time points through POD 21 or DD (fig. 2).

In addition to 11 subjects (3 in the PFC group, 8 controls) who received PAD units, some patients in each group received other blood components intraoperatively: fresh frozen plasma (12.8% in the PFC group and 12.5% in the control group), cryoprecipitate (1 patient in the PFC group), and platelets (7 in PFC group, 7 in the control group; 300 ml for each subject in each of the groups). The number of patients avoiding all types of transfusions of blood and blood components was significantly higher in the PFC group at 24 h, but the difference was not significant at DD (table 4). In contrast, in the protocol-defined target population (blood loss ≥ 20 ml/kg), the percentage of patients avoiding transfusion of any blood and blood components remained significantly greater until DD (table 4).

Platelet counts were significantly lower, by approximately 15–25% on PODs 1–3, in the PFC versus  control groups (P < 0.01;fig. 3). By POD 7, platelet counts were still lower in the PFC group, but both groups had recovered to normal screening levels.

As expected, a high number of AEs were reported in this study in patients undergoing major surgery with substantial (≥ 20 ml/kg) blood loss. The incidence of AEs (table 5) was similar in the PFC group (86%) compared with controls (81%). More serious AEs were reported in the PFC group than in controls (32%vs.  21%;P < 0.05;table 5). However, only the category “Digestive System” was significantly different from controls, mostly because of a higher reported occurrence of postoperative ileus. Four serious AE cases of ileus were reported in the PFC group (vs.  none in controls), one after a rectum amputation, two after major gynecologic tumor excision, and one after radical cystectomy.

The independent data safety monitoring board noted some group imbalances in certain AEs but concluded that there was no clinically consistent pattern or significance. They also concluded that since the investigators were not blinded to treatment allocation, there was a possibility that the Hawthorne effect 15,16may have influenced reporting. Overall mortality was 3% (PFC, 4%vs.  controls, 2%); the difference between groups was not statistically significant. Tumor progression, sepsis, multiorgan failure, and cardiopulmonary complications were responsible for the deaths, and all were considered by the investigators to be a result of underlying disease or condition and to be unrelated to the study drug.

This pivotal phase 3 study has demonstrated that transfusion requirements during high-blood-loss noncardiac surgery can be decreased by using PFC as an intravenous oxygen carrier to enable more extensive preoperative harvesting of autologous blood during hemodilution.

In the intent-to-treat population, PFC-treated patients required fewer erythrocyte units compared with controls in the initial 24 h, but erythrocyte transfusion rates were similar at hospital discharge (table 3). In contrast, overall reduction and avoidance of erythrocyte transfusions were significantly greater in PFC-treated patients in the protocol-defined subgroup of patients experiencing blood loss of 20 ml/kg or greater, and this benefit persisted through hospital discharge. This outcome agrees with mathematical modeling predictions that blood-sparing efficacy of (augmented) ANH can only be achieved when there is high surgical blood loss. 17–19The efficacy of ANH augmented by PFC is derived from the hemodilution itself, based on the fact that more dilute blood is being lost during surgical blood loss, which minimizes loss of erythrocytes. Concomitant intraoperative treatment with PFC and high Fio²provides improved oxygenation, which in turn permits lower hemoglobin concentrations to be safely tolerated.

Since no significant difference in erythrocyte transfusions was noted after POD 1 in the intent-to-treat population (which included 162 subjects who had < 20 ml/kg blood loss), additional analyses were performed in a post hoc  analysis to identify the blood loss level where using augmented ANH with PFC resulted in a significant reduction in transfusion. These analyses demonstated that significant savings were still observed at an estimated blood loss of 10 ml/kg or greater (fig. 2C), which represents 86% (n = 424) of all subjects in the study. For a large number of major noncardiac operations, ANH augmented by using PFC as a temporary intravascular oxygen carrier may therefore represent a new autologous method to decrease allogeneic erythrocyte transfusions.

Regarding safety, a high incidence of AEs was observed in this study (table 5), as expected during major noncardiac surgery, 20–22since two thirds of the patients were undergoing extensive surgical procedures for malignancies (table 2). The higher rate of serious AEs reported in the category “Digestive System” was mostly due to postoperative ileus. A 2% incidence of ileus is not unusual for these operations, but, surprisingly, no ileus was reported in the control group. Reports of sepsis and infection are also of interest, since emulsion particles are initially cleared from the circulation by the phagocytic cells (macrophages in the spleen and Kupffer cells in the liver) of the reticuloendothelial system, 23which could theoretically increase susceptibility to postoperative infections. However, aggregates of all infectious complications were similar in both groups (32%), supporting earlier clinical findings that immune function is not compromised by PFC. 24Overall mortality at 3 months was 3%, which is consistent with a recent report in a similar surgical population. 20Tumor progression, sepsis, multiorgan failure, and cardiopulmonary complications were responsible for these deaths, and no deaths were considered to be related to drug treatment.

Overall, the incidence of AEs was equal in both groups (table 5), although more were reported in the PFC group for the categories “Cardiovascular System” (hypertension) and “Digestive System” (ileus). Hypertension occurred mainly postoperatively and was likely related to mandatory retransfusion of all remaining ANH blood. The clinical consequences in the ileus cases were minor, and no general pattern or pathophysiological mechanisms were found in post hoc  analyses that could link these events to treatment.

The transient decrease in platelet counts due to splenic sequestration and clearance occurring a few days after surgery was expected based on previous phase 2 clinical data. 14Since the number of platelet transfusions and the incidence of bleeding events were similar in both groups, it would appear that this moderate decrease in postoperative platelet count is of little clinical relevance. In addition, the difference in the number of erythrocyte units transfused between the groups persisted (table 3), and the hemoglobin was similar throughout the postoperative period (fig. 1). This indicates that the lower platelet count in the PFC group was not associated with any increased transfusion requirements. This is consistent with earlier studies in human volunteers demonstrating that PFC does not compromise platelet function or increase bleeding time. 25 

The question remains why augmented ANH with PFC successfully reduced erythrocyte transfusions but was associated with a somewhat higher incidence of AEs. It is possible that several centers did not have sufficient previous experience with the degree of ANH and the lower intraoperative hemoglobin concentrations mandated by the protocol. A potential reporting bias (since the study was not double-blinded) to ascribe untoward postoperative events to a new procedure (augmented ANH) using an experimental drug (PFC), as well as a lack of experience with either, may explain some of the increased AE reporting in the PFC-treated group. While the principles of ANH appear simple, there are fluid management and blood volume issues that deserve special attention: hypovolemia must be prevented during blood harvesting and throughout the operation, and hypervolemia must be avoided during retransfusion of ANH units. Performing ANH involves considerable fluid and electrolyte shifts that have to be controlled, and this is where previous experience is helpful to ensure euvolemia. Centers without previous experience may have therefore performed suboptimal ANH and had more problems maintaining euvolumia throughout the procedure, and thus may have encountered more AEs. The data safety monitoring board noted a trend (statistically nonsignificant) for the PFC group to receive smaller volumes of both colloids and crystalloids, although one might have expected the PFC group to receive more fluids intraoperatively because of the 500-ml greater estimated blood loss. This may be an important contributor to some of the AEs reported, because hypovolemia is known to be associated with adverse postoperative outcomes from major surgery. 26There was also a trend (P = 0.063) for the PFC group to have less patients with American Society of Anesthesiologists physical status I and more with status II compared with the control group (table 1), which could potentially be associated with greater postoperative morbidity.

For the conduct of this study, the preoperative ANH procedure delayed surgery by approximately 45–60 min, since it was mandated by protocol to complete the ANH prior to skin incision. In future clinical practice, however, it would only be important to perform ANH prior to major blood loss; thus, it could easily be completed during the first hour of major tumor operations, since the major bleeding phase in most of these operations is not immediately after incision. The future clinical use of artificial oxygen carriers following ANH is therefore not necessarily associated with extra time requirements and would not have to delay the start of surgery.

Postoperative hemoglobin values were slightly higher (∼9.5 g/dl) than the protocol specified hemoglobin-based transfusion trigger (8.5 ± 0.5 g/dl;fig. 1). At each time point only a certain percentage of patients had a hemoglobin concentration less than 8.5 ± 0.5 g/dl and thus would be transfused, while the other patients had hemoglobin values greater than 8.5 ± 0.5 g/dl. Since hemoglobin was measured frequently in the early postoperative period, transfused patients were likely to have a hemoglobin value greater than or close to 8.5 ± 0.5 g/dl in subsequent measurements. Group mean hemoglobin values would therefore be expected to be higher than the protocol-specified postoperative hemoglobin transfusion trigger. More importantly, the postoperative hemoglobin concentrations were similar in both groups. Hence, the different transfusion needs between groups is not related to different transfusion strategies postoperatively, but rather to the lesser intraoperative transfusion needs in the PFC group.

The single-blinded nature of this study deserves comment. Intraoperative blinding was essentially impossible, given the fact that the patients had to be treated differently in the PFC and control groups (preoperative ANH, different intraoperative hemoglobin transfusion triggers). It is therefore difficult to determine whether the estimation of blood loss was inaccurate or potentially biased, or whether blood loss was actually somewhat greater in the PFC group undergoing preoperative ANH, which is associated with vasodilation, decreased blood viscosity, and eventually a reduced margination of platelets at low hematocrit. 27,28Also, AE reporting might have been biased or influenced by the Hawthorne effect. 15,16Finally, the investigators responsible for monitoring postoperative hemoglobin concentrations, to ensure that protocol-mandated transfusion requirements were strictly followed, were not blinded. This might introduce bias as well. However, this study used prospectively defined postoperative hemoglobin transfusion triggers, and the protocol was generally well followed, which resulted in nearly identical postoperative hemoglobin concentrations in both groups (fig. 1). Hence, the potential for bias to influence the efficacy (i.e. , erythrocyte transfusion) outcome of the study appears to be minor.

In the current study, efficacy was greatest when estimated surgical blood loss was 20 ml/kg or greater, but significant transfusion avoidance was also achieved in a broader subset of patients undergoing procedures with a blood loss of at least 10 ml/kg. If blood loss is lower than that, there is no real need to transfuse erythrocytes, and, therefore, there is no chance to improve on the transfusion outcome with any intervention. For this reason, the intent-to-treat analysis was compromised in its ability to demonstrate significant avoidance through discharge, since 33% of the patients bled less than 20 ml/kg and 14% lost less than 10 ml/kg. The efficacy findings in this phase 3 study suggest that the use of PFC as an intravenous oxygen therapeutic to augment autologous blood harvesting may represent a new alternative for the growing number of patients seeking to avoid or minimize the risks of allogeneic erythrocyte transfusions in high-blood-loss elective surgery.

The authors thank the patients who participated in this study and all surgeons, anesthesia and intensive care personnel, statisticians, and clinical research support staff who supported this study with enthusiasm and participated in the study conduct and data collection.

TABLE 

In large phase 2 studies in hemodiluted surgical patients 14, it was demonstrated that, after administration of PFC, patients with low hemoglobin concentrations were improved physiologically (based on mean arterial pressure, heart rate, cardiac output, electrocardiogram changes, and mixed venous oxygen tension). In these studies, subjects underwent preoperative ANH to a hemoglobin concentration of 9.0 g/dl. When the first physiologic transfusion trigger (protocol-defined changes in mean arterial pressure, heart rate, cardiac output, mixed venous oxygen tension, electrocardiogram) occurred during surgery, subjects were randomized to one of four groups: PFC at one of two doses, 1.8 or 0.9 g PFC/kg with Fio²of 1.0, colloid plus Fio²of 1.0, or autologous blood plus Fio²of 0.4. The primary endpoints were an assessment of the efficacy in reversing the physiologic triggers and the duration of the reversal (i.e. , the period of time before a second physiologic trigger occurred, at which time all subjects received autologous blood). These studies thus provide a means of assessing the activity and efficacy of PFC by measuring its ability to reverse and maintain the reversal of the physiologic transfusion triggers during a measured decline in hemoglobin.

A computer model was developed to provide a means of quantitating the added oxygen off-loading capacity provided by the PFC as a “hemoglobin equivalent” value. The percent contribution of the PFC-carried oxygen to the total body oxygen consumption (V̇o²) is expressed in terms of the percent contribution of 1 g/dl erythrocyte hemoglobin to the total V̇o². For instance, if a 1-g/kg dose of PFC provides 10% of V̇o²and each 1-g/dl of erythrocyte hemoglobin provides 5% of the V̇o², then the hemoglobin equivalent is 10/5 = 2 g/dl. The “effective hemoglobin” can then be considered the sum of the erythrocyte hemoglobin and the hemoglobin equivalent. For example, if the hemoglobin equivalent of a dose of PFC is 2 g/dl and the erythrocyte hemoglobin is 8 g/dl, the effective hemoglobin would be 8 g/dl plus 2 g/dl = 10 g/dl.

Analysis of the data from the two phase 2 clinical studies with PFC indicated that the 1.8- and 0.9-g/kg doses of PFC yielded hemoglobin equivalence values of approximately 2.7 g/dl and 1.4 g/dl, respectively, at the time the dose was administered. These data support the conclusion that the intended clinical dose of PFC, 2.7 g/kg, can provide an initial mean hemoglobin equivalent on the order of 4 g/dl at the time of dosing. With simultaneous surgical bleeding and clearance of PFC from the blood by the RES, the analysis indicates that sufficient PFC will remain to provide a hemoglobin equivalent of at least 2.5 g/dl at the time the circulating erythrocyte hemoglobin decreases to 5.5 g/dl. In combination with the available erythrocyte hemoglobin, the remaining effective hemoglobin will be at least 8.0 g/dl (which matched the intraoperative target hemoglobin concentration for transfusing blood in the control group).

In the phase 2 studies, PFC was administered when any one of several predefined physiologic “transfusion triggers” was reached during surgery. After the dose had been administered (during ongoing surgical blood loss), a set of hemodynamic and oxygenation measurements was taken to assess trigger reversal. By knowing the cardiac output, arterial and mixed venous blood gases, body temperature, hemoglobin, and the PFC concentration in the blood at any given time point, it was possible to calculate the contribution of both hemoglobin and PFC to total body V̇o². Using data for the 73 subjects for whom full data sets were available, the projected mean hemoglobin equivalent of a 2.7-g/kg dose of PFC was determined to be 2.4 ± 1.7 g/dl, by extrapolation from hemoglobin equivalent values calculated for the patients receiving 1.8- and 0.9-g/kg PFC doses.

This prediction of hemoglobin equivalence actually underestimates the potential hemoglobin equivalence because, at the time the blood samples for PFC concentration were obtained, circulating perflubron concentrations had already been reduced from the immediate postdosing levels due to ongoing blood loss and clearance by the RES. Also, the time elapsed before the postdosing sample was collected varied from patient to patient. Hence, a more relevant estimate of the total initial hemoglobin equivalence can be obtained by calculating the PFC concentration based on each subject's estimated blood volume and the known total dose of PFC administered. Using this approach, the mean predicted hemoglobin equivalence for a total 2.7-g/kg dose of PFC was found to be 4.0 ± 2.6 g/dl.

Removal of PFC from the circulation by the RES exhibits dose-dependent pharmacokinetics, and phase 1 studies with PFC demonstrated that the half-life of PFC in the blood is related to the absolute dose (i.e. , total grams of PFC) administered. 25Its clearance from the circulation can therefore be predicted, as can its removal from the circulation as surgical bleeding occurs. It is therefore possible to estimate the hemoglobin equivalent adjusted for RES clearance and surgical blood loss at any point during the surgical procedure.

Extensive calculations of hemoglobin equivalence were performed for the proposed dosing regimen of administering a 1.8-g/kg dose of PFC at a erythrocyte hemoglobin concentration of 8.0 g/dl, with a second dose of 0.9 g/kg PFC given at an erythrocyte hemoglobin concentration of 6.5 g/dl. These calculations were performed at differing rates of blood loss for patients with different blood volumes and indicated that it was safe under all circumstances to allow the PFC-treated patient's erythrocyte hemoglobin to decrease to the point at which erythrocyte transfusion was mandated in the protocol, i.e. , 5.5 g/dl, as the effective hemoglobin remained greater than 8.0 g/dl.

Verification of the accuracy of the model for calculation of the hemoglobin equivalence has been accomplished by comparison to experimental clinical data from the phase 2 studies. In the 48 subjects for whom data were available, the hemoglobin equivalence at the second postdosing transfusion trigger, calculated from the clinical data, was then compared with the hemoglobin equivalence calculated by the model. There were no significant differences in hemoglobin equivalence (model vs.  clinical); the mean difference was −0.2 ± 0.7 g/dl, and there was a highly significant correlation between the two variables (R²= 0.90). Hence, the model appears to incorporate all of the critical parameters necessary to produce acceptable predictions of hemoglobin equivalency. This model was therefore used to design the current study and to ensure that the PFC-treated patients could safely be taken to the low hemoglobin concentrations that were targeted in this study.