Reduction of Blood Loss and Transfusion Requirement by Aprotinin in Posterior Lumbar Spine Fusion

Abstract

Aprotinin reduces blood loss in many orthopedic procedures. In posterior lumbar spine fusion, blood loss results primarily from large vein bleeding and also occurs after the wound is closed. Seventy-two patients undergoing posterior lumbar spine fusion were randomly assigned to large-dose aprotinin therapy or placebo. All patients donated three units of packed red blood cells (RBCs) preoperatively. Postoperative blood loss was harvested from the surgical wound in patients undergoing two- and/or three-level fusion for reinfusion. The target hematocrit for RBC transfusion was 26% if tolerated. Total (intraoperative and 24 h postoperative) blood loss, transfusion requirements, and percentage of transfused patients per treatment group were significantly smaller in the aprotinin group than in the placebo group (1935 ± 873 vs 2809 ± 973 mL per patient [P = 0.007]; 42 vs 95 packed RBCs per group [P = 0.001]; 40% vs 81% per group [P = 0.02]). Hematological assessments showed an identically significant (a) intraoperative increase in both thrombin-antithrombin III complexes (TAT) and in activated factor XII (XIIa) and (b) decrease in activated factor VII (VIIa), indicating a similar significant effect on coagulation in patients of both groups (P = 0.9 for intergroup comparisons of postoperative VIIa, XIIa, and TAT). Intraoperative activation of fibrinolysis was significantly less pronounced in the aprotinin group than in the placebo group (P < 0.0001 for intergroup comparison of postoperative D-dimer levels). No adverse drug effects (circulatory disturbances, deep venous thrombosis, alteration of serum creatinine) were detected. Although administered intraoperatively, aprotinin treatment dramatically reduced intraoperative and 24-h postoperative blood loss and autologous transfusion requirements but did not change homologous transfusion in posterior lumbar spine fusion.

Implications 

In our study, aprotinin therapy significantly decreased autologous, but not homologous, transfusion requirements in posterior lumbar spine fusion.

The extensive use of pre-, intra-, and postoperative red blood cell (RBC) salvage and the tolerance of low hemoglobin levels are highly effective in reducing blood loss and transfusion requirements in spine surgery (^1–3). However, spine surgery still carries the risk of significant blood loss, which often complicates the surgical approach and may still require allogeneic blood transfusion (¹). Furthermore, in spine surgery, bleeding has some specificities: it results primarily from large vein bleeding and persists after the wound is closed (¹). Therefore, additional measures to reduce bleeding and transfusion requirement must be considered in spine surgery. Aprotinin reduces intraoperative bleeding and transfusion requirements in cardiac surgery (⁴), liver resection (⁵), and orthopedic surgery (^6,7). The mechanism of this blood-sparing effect is unclear (⁸), although a protective effect on platelet membrane binding function and the inhibition of intraoperative fibrinolysis have been suggested (⁸). The current study is a prospective, double-blinded, randomized, controlled trial of the effect of aprotinin on both perioperative blood loss and packed RBC requirements in patients undergoing elective lumbar interbody fusion through a posterior approach (¹).

In addition, because there is some concern that, as an adverse effect of its antifibrinolytic effect, aprotinin could increase the risk of thrombosis (⁹), activated factor XII (XIIa), activated factor VII (VIIa), and prothrombin fragment F1 + 2 (F1 + 2), early markers of coagulation activation, were measured intra- and postoperatively (¹⁰).

Methods

Our institutional review board approved the protocol, and all patients provided their written informed consent. Between October 1996 and November 1998, adult patients scheduled for elective lumbar interbody spine fusion through a posterior approach (¹) were assigned in a double-blinded fashion, by using a computer-generated random code, to receive either intraoperative aprotinin or the equivalent volume of placebo. Surgery performed for degenerative disease was exclusively considered for this trial. Patients were distributed among three subgroups (one, two, or three spine levels fused). Randomization was both stratified by the number of fused levels and blocked in groups of four before the induction of anesthesia. Patients were excluded if they had any suspected bleeding disorder—inherited bleeding disorder, aspirin or dipyridamole use within 10 days before the operation, anticoagulation therapy that could not be interrupted, preoperative fibrinolysis requiring an intraoperative antifibrinolytic treatment, preoperative platelet count <150,000/mm3, prothrombin time (PT) (expressed as percentage of prothrombin complex in comparison with healthy control) <80%, or prolonged activated partial thromboplastin time; previous venous or arterial thrombosis or any biological abnormality likely to induce thrombosis; had a known allergy to aprotinin or a possible previous exposure to the drug; were pregnant; had impaired renal function defined as serum creatinine level >150 moles/L; had any clinical or biological sign of liver impairment; had any contraindication to predepositing autologous blood or if it was impossible; or were <18 yr of age.

The aprotinin group received an initial dose of 2 × 106 KIU of aprotinin over a 20-min period after the induction of anesthesia, followed by a continuous infusion of 5 × 105 KIU/h administered via an infusion pump until skin closure (^5–7,11). An additional bolus of 5 × 105 KIU of aprotinin was infused every 3 RBCs. Each patient in the control group received equivalent volumes of placebo (0.9% saline solution) at the respective times. Aprotinin (Trasylol®; Bayer Laboratory, Puteaux, France) was supplied in bottles containing 5 × 105 KIU in 50 mL of saline without preservatives or additives. A venous line was used only for aprotinin/placebo infusion. Anesthetic management and intraoperative care were standardized throughout the study period. The same limited team cared for patients during the study.

Subcutaneous heparin (5000 IU, Fragmine®; Pharmacia SA, Guyancourt, France) administration was begun 12 h before surgery until discharge for prophylaxis of deep vein thrombosis. After oral premedication with hydroxyzine (2 mg/kg), general anesthesia was induced with IV thiopental (4–6 mg/kg), sufentanil (0.3–0.5 μg/kg), and vecuronium (0.1 mg/kg), followed by endotracheal intubation. Anesthesia was maintained with a continuous IV infusion of 0.1–0.3 μg · kg⁻¹ · h⁻¹ sufentanil, isoflurane, and boluses of 1 mg of vecuronium as required. The lungs were ventilated mechanically. Arterial blood pressure was maintained within 20% of preinduction values by adjusting the isoflurane concentration and/or by intravascular fluid administration. Esophageal temperature was maintained >36°C. Lactated Ringer’s solution was infused intraoperatively at a basal rate of 5 mL · kg⁻¹ · h⁻¹. An indwelling catheter was placed in a radial artery for arterial pressure monitoring and blood sampling.

During surgery, patients were on a prone sitting frame with the abdomen hanging free. Arthrolaminectomy allowed the spinal canal approach for surgery. Spine fusion included the placement of a cage filled with spongious bone in the interbody space (Oarlock® interbody fusion cage; Biomat Laboratory, Igny, France), bone placement between the transverse processes, and vertebra bodies fixation with two plates screwed into the vertebra pedicles. One, two, or three spine levels were fused. Graft bone was obtained from the ilium. At the end of surgery, patients undergoing two and/or three fused levels had two suction drains inserted into the wound before closure for reinfusion of harvested shed blood. No cell saver was used.

Intraoperative blood loss was carefully measured by adding the volume of blood in suction bottles and the weight of sponges. All fluids added to the surgical field intraoperatively were carefully quantified and deducted from the measured blood loss. The anesthetist responsible for intraoperative patient management, who was unaware of the treatment regimen, collected the intraoperative blood loss data. Shed blood harvested up to 6 h postoperatively was systematically reinfused. Reinfused blood was not washed. Drainage that occurred after 6 h postoperatively was not returned to the patient and was included in 24-h postoperative blood loss final assessment.

All patients deposited preoperatively three units of autologous RBCs. Uniform transfusion criteria were adhered to in all patients during the intra- and postoperative periods. Intraoperative hematocrit was assessed systematically every 1 h and more often if necessary at the discretion of the anesthetist in charge of the patient. Postoperative hematocrit was assessed 6, 12, 18, and 24 h after surgery and more often if necessary to conduct postoperative packed RBC transfusion. The target hematocrit value for intra- and postoperative RBC transfusion was 26%, except in patients >60 yr old, those with preexisting heart or lung disease, or if this low hematocrit level was not clinically tolerated (⁶). Hydroxyethyl starch was administered for volume replacement up to 3.3% of total infused volume. Thereafter, the protocol called for saline or fresh-frozen plasma to be infused, if necessary.

After surgery, surgeons were subjectively asked to quantify intraoperative bleeding on a three-point scale (1 = minimal bleeding, 2 = usual bleeding, 3 = severe bleeding).

Blood samples were collected from each patient after the induction of anesthesia at the following times: before skin incision and before the administration of aprotinin or placebo (T1), at the end of surgery (T2), and 24 h after the infusion of aprotinin or placebo (T3). Blood drawn from patients was mixed with one-tenth volume of 0.129 M trisodium citrate. After two centrifugations at 2400 g for 20 min at 4°C, plasma was either immediately analyzed or frozen in dry ice in small aliquots and stored at −80°C. Reference pool plasma consisting of 25 plasma samples from healthy controls (men or women not taking oral contraceptives, aged 20–40 yr) was also stored at −80°C. All blood samples were tested for the following variables: hematocrit and platelet count and hemostatic tests: fibrinogen, D-dimers (normal levels <400 ng/mL), thrombin-antithrombin III complexes (TAT) (normal levels <20 ng/mL), XIIa (normal levels <3.08 ± 2), and F1 + 2 (normal levels <1.1 ± 0.3) were assessed by using an enzyme-linked immunosorbent assay (Diagnostica Stago, Asnieres, France or Shield Diagnostics Ltd, Dundy, UK). VIIa (normal <4.26 ± 2 ng/mL) was measured by using the VIIa-rTF assay (Diagnostica Stago).

Any circulatory disturbance possibly related to drug intolerance was noted. Deep vein thrombosis was assessed daily by clinical examination until discharge. Patients with clinical symptoms of deep vein thrombosis were scheduled to undergo lower limb venography. Serum creatinine levels were assessed preoperatively, 24 h after surgery, and 3 days postoperatively.

The primary end point of the study was perioperative blood loss (summation of intraoperative and 24-h postoperative blood loss). Perioperative blood loss had been estimated as being 2200 ± 1000 mL in posterior lumbar interbody spine fusion using a cage in our institution in 40 patients during the year preceding the present study. To detect a 30% reduction in blood loss with 80% power and an α risk of 0.05 in an unilateral hypothesis power analysis indicated that 72 patients would have to be studied (¹²). Qualitative variables were analyzed using the χ2 test. Intraoperative, postoperative, and total perioperative blood loss values were subjected to two-way analysis of variance taking into account age, duration of surgery, weight, number of fused levels, and repeat surgery. Intergroup comparisons of the amount of blood collected postoperatively for reinfusion, packed RBCs transfused, and biological variables were analyzed using the nonparametric Mann-Whitney U-test. The temporal evolution of hemostatic variables was evaluated using a variance analysis. If a total time effect was significant, the different times were compared with the preoperative value by using Wilcoxon’s test. The Bonferroni correction was applied to avoid the statistical error induced by the multiplication of the tests. Results are expressed as means ± SD or median (ranges) as appropriate. P < 0.05 was the minimal level of significance. In addition, a stepwise logistic regression model was used to assess whether aprotinin treatment was independently correlated with blood loss when the confounding variables (number of fused levels, duration of surgery, repeat surgery, age, sex ratio, weight) were taken into account. In this model, a P value < 0.1 was considered significant (¹³). The aim of this multivariate analysis, which included all patients, was to assess factors likely to explain blood loss variations. The dependent variable was blood loss. Independent variables were number of fused levels, duration of surgery, repeat surgery, age, sex ratio, weight.

Results

Seventy-two consecutive patients were included in this study. Patient demographic and intraoperative characteristics, number of fused spine levels, and frequency of repeat surgery did not differ between the groups (Table 1).

Table 1: Patient Characteristics and Intraoperative Data

Intergroup comparisons show a significant overall reduction of total perioperative and 24-h postoperative blood loss in the aprotinin group compared with the placebo group (Table 2). Intraoperative blood loss and the amount of blood collected postoperatively for reinfusion tended to be larger in the placebo group than in the aprotinin group, although the difference did not reach significance. In addition, our results show a significant independent association between perioperative blood loss and both the treatment regimen and the amount of fused levels, with no significant interaction between these two main factors. In the stepwise logistic regression analysis, aprotinin treatment was significantly correlated with blood loss after adjusting for number of fused levels, duration of surgery, repeat surgery, age, sex ratio, and weight (P = 0.0001). In the stepwise logistic regression analysis, three variables were independently correlated with blood loss: treatment (P = 0.0001), age (P = 0.07), and duration of surgery (P = 0.01). Four variables were not correlated with blood loss: number of fused levels (P = 0.8), sex ratio (P = 0.2), weight (P = 0.5), and repeat surgery (P = 0.2).

Table 2: Blood Loss and Transfusion Requirements

The number of perioperative packed RBCs transfused and the percentage of transfused patients per treatment group were significantly lower in the aprotinin group than in the placebo group. Aprotinin significantly reduced perioperative autologous RBC requirements but did not significantly affect perioperative homologous RBC requirements. One patient in the placebo group and none in the aprotinin group received more than 5 units of RBCs. No fresh-frozen plasma was transfused.

Subjective bleeding, as assessment by the surgeon and expressed in Table 3, shows a significantly lower bleeding score in the aprotinin group.

Table 3: Subjective Surgical Bleeding Evaluation

Preoperative hematological variables (T1) did not differ significantly between the aprotinin and placebo groups (Table 4). Intraoperative fibrinolysis, as assessed by intergroup comparison of postoperative fibrinogen and D-dimer levels (T2), was significantly decreased by aprotinin therapy. This antifibrinolytic effect did not persist 24 h later (T3). Intraoperative activation of coagulation did not differ significantly between the treatment groups. XIIa, F1 + 2, and TAT increased significantly, whereas VIIa decreased significantly in both treatment groups at the end of surgery (T2).

Table 4: Hematologic Variables

No adverse effect believed to be related to the treatment was recorded during the follow-up period. No patient had any clinical sign suggestive of venous thrombosis. No postoperative increase of serum creatinine was observed (P = 0.9 for intergroup comparisons of plasma creatinine levels 24 h and 3 days postoperatively).

Discussion

Our results show that, in patients undergoing posterior lumbar spine fusion, aprotinin significantly reduced total intraoperative and 24-hour postoperative blood loss, packed RBC requirement, and the number of transfused patients. Furthermore, the subjective assessment of intraoperative bleeding by the surgeon was better in the aprotinin group. Our findings are in line with previous studies conducted under various surgical conditions demonstrating a significant reduction in intraoperative and postoperative blood loss with a subsequent decrease in transfusion requirement with aprotinin (^4–7).

Posterior lumbar interbody spine fusion warranted specific evaluation of the blood-sparing effect of aprotinin. In this procedure, intraoperative blood loss results primarily from direct large internal vertebral venous plexus, which cover the spinal canal floor (¹⁴). In posterior lumbar spine fusion, the largest blood loss may occur after the wound is closed (¹), which is why, in the present evaluation, criteria for aprotinin efficacy were total, intraoperative, and 24-hour postoperative blood loss and transfusion requirements. Consequently, whether intraoperative aprotinin could reduce surgical blood loss partly related to direct vessel bleeding and remain efficient for a 24-hour period was questionable. Furthermore, maneuvers for hemostasis of bleeding from the vertebra bodies are technically demanding and may increase nerve root exposure to surgical hazards (^1,14).

We controlled some factors shown to influence blood loss and transfusion requirements in posterior lumbar spine fusion (^15,16): operating time, repeat surgery, number of fused vertebrae, body weight, duration of postoperative suction drainage, and number of fused levels. Attention was focused to adequately protect pressure points to minimize central venous pressure alteration because high pressure in epidural veins causes surgical bleeding (¹⁷).

Other factors are unlikely to have altered this trial. Any fluid added in the operative field for surgical needs was carefully quantified. Standard rules accounted for transfusion criteria and for appreciation of anemia intolerance. Pre- and postoperative hematocrit levels did not differ significantly between treatment groups, which suggests a similar trigger for transfusion in each group. Intergroup comparisons of intraoperative and six-hour postoperative blood loss did not reach significance, but aprotinin therapy significantly decreased both 24-hour postoperative and total perioperative blood loss compared with placebo. Indeed, in lumbar spine fusion, late intraoperative bleeding is likely to be shed into suction drains in the early postoperative period, making it impossible to discriminate between bleeding occurring during these two periods (¹). Furthermore, the beneficial effect of aprotinin was longer than aprotinin’s half-life, which is approximately 150 minutes. This is in accordance with the fact that the pharmacologic effects of a drug may last longer than its half-life.

The optimal dose of aprotinin therapy has never been assessed. Our dose regimen was chosen because it consistently reduced blood loss and transfusion requirements in previous studies conducted during extracorporeal bypass (^5–7,11).

The mechanism by which aprotinin decreases intraoperative bleeding is not fully elucidated (⁸), and our data do not allow any conclusion in this regard. Our results demonstrate the inhibition of intraoperative fibrinolysis by aprotinin therapy, but platelet function was not assessed. Inhibition of fibrinolysis by aprotinin has been consistently demonstrated during cardiopulmonary bypass (¹⁸). An early protective effect of aprotinin on platelet membrane binding function, altered by the contact of blood components with the surface of the extracorporeal bypass, has also been previously advocated (¹⁹). In this regard, von Willebrand factor and/or platelet membrane glycoproteins could be preserved (¹⁹). However, this beneficial effect on platelet function in cardiac surgery has been questioned (²⁰). The reduction of an overall inflammatory reaction by aprotinin has never been associated with any reduction in blood loss (²¹). Topical aprotinin also significantly reduced blood loss and transfusion requirements in cardiac surgery, although no systemic aprotinin was detectable (²²). In orthopedic surgery, data are also controversial (⁸). In hip surgery, a reduction in blood loss with large-dose aprotinin occurred without any associated antifibrinolytic effect (^6,7). The effect of aprotinin on platelet function was also assessed in hip surgery in two studies that demonstrated a decrease in blood loss with aprotinin (^7,11). Janssens et al. (⁷) did not find any change in platelet count, bleeding time, in vivo intraoperative platelet activation, thromboglobulin level, and in vitro intraoperative induced aggregability with aprotinin therapy. Conversely, Haas et al. (¹¹) showed that aprotinin preserved spontaneous and ADP-induced platelet aggregation, preserved adhesivity, and decreased postoperative platelet aggregates.

No adverse effect of aprotinin treatment was shown in the present study. No thromboembolic event occurred in either group. This concern is substantiated by anecdotal reports of rare cases of pulmonary embolism during liver transplantation (²³), suspicion of graft patency impairment in a study conducted in a small population (²⁴), or induced thrombosis recorded in a laboratory animal with a baseline PT 500 times higher than in humans (⁹). Data collected in orthopedic surgery (^6,7), cardiac surgery (⁴), and abdominal surgery (⁵) do not support this concern. A trend toward a less frequent incidence of vein thrombosis associated with aprotinin administration has even been suggested in orthopedic surgery (⁶). In the present study, in vivo biological assessments support a safety trend that is in contrast with in vitro studies (²⁵). Indeed, VIIa, XIIa and F1 + 2 changes are early indicators of the biochemical silent phase of coagulation activation (¹⁰). These factors are prominent in situations of increased risk of arterial and venous thrombosis, such as smoking, heart disease, cancer, and pregnancy (^10,26). In the present trial, aprotinin therapy did not alter their evolution. Nevertheless, in our investigation, deep vein thrombosis was not detected with an objective diagnostic test in an overall surveillance program in which it could have been higher. No allergic reaction to the drug occurred during the study. However, no patient had received aprotinin previously. Indeed, allergic reactions are a potential risk, mainly in cases of reexposure (²⁷), because the prophylactic use of aprotinin is increasing, and aprotinin is a bovine-derived drug. Finally, serum creatinine variation was similar in both treatment groups at a three-day follow-up. This does not rule out a transient dysfunction not detected by the investigated variables (⁴).

In the present study, aprotinin therapy significantly reduced only autologous transfusion and had only an insignificant influence on allogeneic transfusion. All patients but one received five or fewer units of RBCs in the placebo group during the follow-up period. Restrictive transfusion criteria and the blood salvaging procedures applied during this trial are likely to account for this low transfusion rate. Indeed, a significant blood-sparing effect occurs with packed RBC predonation (¹), postoperative reinfusion of shed blood harvested from the lumbar wound (²⁸), and a low hematocrit trigger for blood transfusion (²). Neither isovolemic hemodilution nor induced low blood pressure were used because the blood-sparing efficacy of these procedures is doubtful (^29,30). However, additional blood salvaging techniques could have further reduced allogeneic transfusions (^29,30). Intraoperative blood salvaging reduces transfusion requirements in spine surgery but is not routinely used in lumbar spine fusion in our institution (¹). RBC apheresis performed twice, 25 ± 5 days before surgery, was reported to allow inexpensive, consistent, and safe storage of five units of RBCs preoperatively in orthopedic surgery (³¹). Because a number of adverse effects may occur, prophylactic aprotinin administration should be restricted to situations in which combined blood salvaging techniques are not available. An additional indication may be anticipation of a difficult surgical approach, a situation in which a dry surgical field may be technically helpful and may increase the safety of nerve root surgical exposure.

In conclusion, the prophylactic aprotinin administration significantly decreases autologous transfusion requirements, but it has no significant effect on homologous blood requirements if posterior lumbar spine fusion is performed under the blood salvaging conditions of this study.