Activated factor VII (FVIIa) plays a key role in hemostasis 1. The effect of FVIIa is based first on its binding to tissue factor exposed on the subendothelium. The FVIIa-tissue factor complex is formed only in the region of endothelial damage, so that a local hemostasis occurs via the subsequent activation of factors IX and X and the formation of thrombin. After that, thrombin activates platelets and the factors V and VIII at the site of injury. Since 1996, a recombinant activated factor VII (rFVIIa) has been available (NovoSeven^®; Novo Nordisk A/S, Bagsvaerd, Denmark). The rFVIIa doses used for registered indications (90 μg/kg body weight at intervals of 2 to 3 hours in hemophilia A or B with inhibitors, in acquired hemophilia, and in Glanzmann thrombasthenia and 15 to 30 μg/kg body weight every 4 to 6 hours for the treatment of bleeding in patients with congenital factor VII deficiency) exceed the physiological FVIIa concentration many times over. This results in an additional pharmacodynamic effect: irrespective of tissue factor as rFVIIa activates factor X directly on the surface of the platelets activated at the site of injury, so that large amounts of thrombin develop locally ('thrombin burst') and eventually induce the formation of fibrin from fibrinogen.

In a literature search, the databases Medline, Biosis, Embase, and Current Contents were screened by using key index terms such as rFVIIa, factor VIIa, recombinant activated factor VII, recombinant blood clotting factor 7a, recombinant FVIIa, or NovoSeven^® (period from 1980 to March 2006). The results were used to select clinical data and case reports on the clinical use of rFVIIa in abdominal surgery (no reviews or abstracts from conferences or scientific meetings). Moreover, the references listed in the literature found were looked through and the manufacturer was asked for publications on the subject matter described. All publications found with these key terms were analyzed as to whether they are about patients undergoing abdominal or vascular surgery, without pre-existing coagulation disorder, and treated with rFVIIa due to perioperative bleeding. Reports on trauma patients or patients with abdominal trauma were excluded. Moreover, all publications were analyzed for assessable information on the target variables 'cessation of bleeding or reduction of blood loss' ('reduction or cessation of bleeding'), 'mortality', and 'occurrence of thromboembolic complications'. These target variables were defined as 'met' (responder) or 'not met' (nonresponder) according to the judgment of the authors of the publications reviewed.

To differentiate case series from case reports, the former had to describe at least three patients being treated with rFVIIa for acute bleeding in one publication, at least one of them having to have undergone surgery in the abdominal region. The results of the descriptive analysis were reported in the form of means and ranges.

A subsequently performed meta-analysis of the case series investigated the mean relative frequency of the effects 'reduction or cessation of bleeding', 'mortality', and 'thromboembolism' of the patients treated with rFVIIa in a random effects model. Analysis was based on the assumption that the selected studies constitute a random selection, whose variance to be considered results from the addition of the individual study portion σ^i=p^i(1−p^i)/ni  MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuqqRPxAKvMB6bYrY9gDLn3AGiuraeXatLxBI9gBaebbnrfifHhDYfgasaacPi6xNi=xH8viVGI8Gi=hEeeu0xXdbba9frFj0xb9qqpG0dXdb9aspeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGacaGaaiaabeqaaeqabiWaaaGcbaacciGaf83WdmNbaKaadaWgaaWcbaGaemyAaKgabeaakiabg2da9iqbdchaWzaajaWaaSbaaSqaaiabdMgaPbqabaGccqGGOaakcqaIXaqmcqGHsislcuWGWbaCgaqcamaaBaaaleaacqWGPbqAaeqaaOGaeiykaKIaei4la8IaemOBa42aaSbaaSqaaiabdMgaPbqabaGccaaMc8oaaa@4149@ and the estimated variance σ^ MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuqqRPxAKvMB6bYrY9gDLn3AGiuraeXatLxBI9gBaebbnrfifHhDYfgasaacPi6xNi=xH8viVGI8Gi=hEeeu0xXdbba9frFj0xb9qqpG0dXdb9aspeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGacaGaaiaabeqaaeqabiWaaaGcbaacciGaf83WdmNbaKaaaaa@2FA8@ between the studies. A mean (weighted) effect along with its 95% confidence interval (CI) was calculated for each trial. In a second step, the placebo-controlled studies on the use of rFVIIa in surgical interventions published in the form of congress abstracts or original articles were evaluated with regard to the thromboembolic risk. Therefore, the mean (weighted) odds ratio of all studies was calculated along with the 95% CI, and the results were visualized using forest plots 7.

The meta-analysis of the present case series on the efficacy (reduction or cessation of bleeding) of rFVIIa in patients undergoing an operation in the field of abdominal surgery yielded an estimated mean effect of 73.2% of patients with a 95% CI of 51.0% to 95.4%. A number of case reports, though subject to a publication bias, document efficacy of rFVIIa by reporting a close temporal relationship between the administration of rFVIIa and cessation of bleeding 203033. These results confirm the experiences of different fields of operative medicine 35, describing the potential of treatment with rFVIIa for the control of serious bleeding in surgical patients.

The rate of cessation or reduction of bleedings after rFVIIa cannot be estimated by the analysis of case reports as these are associated with strong publication bias. Due to a lack of prospective trials, the analysis of case series may yield more reliable results. The mean survival rate of 53.0% determined from nine case series (95% CI 26.4% to 79.7%) demonstrates the poor prognosis of patients with severe postoperative bleeding that could not be controlled with conventional measures. The survival rate of 10.0% in nonresponders to rFVIIa was lower than in patients with reduction or cessation of bleeding after administration of rFVIIa (65.5%; P = 0.003). These results emphasize the relevance of sufficient control of bleeding for the outcome of patients. rFVIIa-refractory bleeding requires immediate measures of bleeding control (reoperation, packing, intraoperative interventional catheter procedures, and so on) to maintain the chance of survival.

Based on more than 10 years of experience with the treatment of patients with hemophilia, the frequency of side effects of rFVIIa can be expected to be less than one side effect per 1,000 standard doses 36. As with any hemostatic drug, special attention should be given to thromboembolic events. Without a control group and due to the known increased thromboembolic risk of surgical procedures, it is not possible to establish an association with rFVIIa on the basis of the calculated mean rate of thromboembolism of 16.5%. Therefore, we performed an additional meta-analysis of placebo-controlled studies, in which the prophylactic use of rFVIIa was investigated in the context of surgical interventions. No significantly increased risk of thromboembolism was found in any of these studies. Our meta-analysis of studies using rFVIIa for therapeutic rather than prophylactic purposes did not show any significantly increased risk of thromboembolism after administration of rFVIIa compared with placebo. The mean estimated risk of thromboembolism of 6.42% in the eight studies on the prophylaxis of bleeding was within the 95% CI of 1.2% to 31.8% determined for the case series of bleeding therapy. Of note, thromboembolic complications were observed when rFVIIa was administered concomitantly with activated prothrombin complex concentrates (aPCCs) 3637. The thromboembolic risk due to the pretreatment with procoagulatory drugs should be taken into consideration in the benefit-risk assessment but per se does not constitute a contraindication for the administration of rFVIIa 35.

Since there are no controlled studies on bleeding therapy in abdominal surgery, the formal degree of evidence on the use of rFVIIa in patients with bleeding during and after operations in the field of abdominal surgery is low and currently corresponds to an expert opinion (degree of evidence C). To safely and effectively tailor the individual treatment of refractory bleeding with coagulation factor concentrates for which no evidence of efficacy can be derived from controlled studies, therapy algorithms considering the severity of bleeding, the patient's general health, and the availability of alternative hemostatic options, and so on can be helpful. Clark and colleagues 9 defined three criteria as a condition for the administration of rFVIIa: (a) transfusions of packed red cells corresponding to at least 1.5-fold the amount of blood volume (>15 units), (b) persistence of severe bleeding in spite of optimal conventional therapy, and (c) no foreseeable immediate surgical bleeding control. The treatment algorithm shown in Figure 3 describes a differentiated therapeutic and diagnostic procedure based on our own clinical experiences as well as on recently published recommendations 35. It would be desirable to further substantiate this model and to provide evidence of the reduction of morbidity and mortality in randomized studies. The following findings are important when using the algorithm: (a) administration that is too late can compromise the success of treatment, (b) therapy with rFVIIa is useless in moribund patients in desolate overall circumstances, and (c) successful treatment of bleeding implies attempts for correction of acidosis (target pH of ≥7.2), fibrinogen (target of ≥100 mg/dL), platelets (target of ≥50,000/μL), and hematocrit (target of >24%) 35 which may also impact the efficacy of rFVIIa.

These findings are also supported by individual case series. The three nonresponders reported by Clark and colleagues 9 had already received between 25 and 44 packed red cell units at the time of rFVIIa administration and developed a serious coagulopathy. Mayo and colleagues 15 found that an improvement of the coagulation status had been achieved in patients responding to rFVIIa in the period between ordering of rFVIIa and administering of rFVIIa, whereas this was not the case in the group of nonresponders. Acidosis is associated with a reduced response to rFVIIa 38, so that its correction is of particular importance. The effect of rFVIIa is less impaired in patients with hypothermia 39. Nevertheless, a body temperature in the normal range is of benefit to bleeding control 3839.

Nowadays, with the budgeting in the health care system, the considerable costs of rFVIIa are an important factor in treatment decision. Therefore, pharmacoeconomic calculations are required which should involve costs of transfusion and intensive care treatment as well as length of stay in the hospital. Based on a population of patients who had received more than 5 units of packed red blood cells, a hypothetic model was able to demonstrate the cost-effectiveness of rFVIIa for administration after more than 14 units of packed red cells 40.

The use of rFVIIa in serious bleeding during or after operations in the field of abdominal surgery has been described as effective and safe in various case reports and series. If there are no pre-existing coagulation disorders, there are currently no indications for the prophylactic administration of rFVIIa. However, rFVIIa can be taken into consideration as an additional therapeutic option if serious bleeding was refractory to conventional treatment. The degree of evidence about this corresponds to that of an expert opinion (degree C). According to the experience gained to date and to the placebo-controlled studies, safety is clinically sufficient, although an increased risk of thromboembolism for patients at risk is suspected. Prospective randomized studies need to investigate the efficacy and cost-effectiveness of rFVIIa in this indication in order to allow a final assessment of the importance of this treatment to be made.

• While all case reports documented recombinant activated factor VII (rFVIIa) to be effective, which represents a substantial publication bias, the meta-analysis of the case series showed a reduction of bleeding in 73.2% of patients.

• Patients who responded to the administration of rFVIIa with reduction or cessation of bleeding had a significantly higher probability of survival.

• Thromboembolic complications occurred in six of 36 patients from the case series (16.5%), which was not increased compared with the meta-analysis of thromboembolic events from the eight placebo-controlled studies.

CI = confidence interval; ERCP = endoscopic retrograde cholangiopancreaticography; FVIIa = activated factor VII; rFVIIa = recombinant activated factor VII.

CvH, SJ, SZ, and JK have received speaker's honoraria from pharmaceutical companies, including Novo Nordisk Pharma GmbH (Mainz, Germany), which are engaged in the field of hemostasis. Furthermore, CvH, SJ, and CS declare that they took part in multicenter studies sponsored by Novo Nordisk Pharma GmbH. DJ declares that his company received honoraria from Novo Nordisk Pharma GmbH for activities as an independent medical advisor, including collaboration in the drafting of this manuscript. K-DW declares that he has no competing interests.

CvH contributed to the conception and design of the work, sampling of data, analysis and interpretation of the data, and drafting of the article. SJ and CS contributed to the conception and design of the work, sampling of data, interpretation of the data, and revision of the article for content. SZ contributed to the conception and design of the work, interpretation of the data, and revision of the article for content. K-DW contributed to the analysis and interpretation of the data, revision of the article for content, and approval of the version to be published. DJ contributed to the conception and design of the work, sampling of data, interpretation of the data, and drafting of the article. JK contributed to the conception of the work, interpretation of the data, revision of the article for content, and approval of the version to be published. All authors read and approved the final manuscript.