In 1999, Health Canada approved the use of recombinant activated factor VII (rFVIIa; Niastase; Novo Nordisk, Mississauga, ON, Canada), a hemostatic drug, for the treatment of bleeding episodes in hemophilia patients with inhibitors to factor VIII or IX.Footnote 1 It was soon found to be useful for controlling bleeding in non-hemophilia patients,1,2 and since approval, it has been increasingly used off-label in various clinical settings, e.g., for refractory bleeding in cardiac surgery.3-5 A recent review of its use in hospitals across the United States found that the off-label use of rFVIIa increased by approximately 140-fold from 2000 to 2008, whereas its use for approved indications increased by only fourfold.3

This off-label use has been the focus of some controversy. On the one hand, although rFVIIa may decrease blood loss and blood transfusion, opponents emphasize that there is no evidence of a mortality benefit and its administration carries thrombotic risks, possibly in a dose-dependent manner. Thus, they have proposed that its off-label use should be confined to clinical trials.6-9 On the other hand, proponents counter that judicious use of rFVIIa outside hemophilia is warranted when the risk-benefit profile is favourable, such as in life-threatening coagulopathy after surgery or trauma.10-15 In either case, the “runaway” increase in the off-label use of rFVIIa in United States hospitals is disconcerting, as it would suggest that the drug is being used outside the select group of patients in whom its benefits may potentially outweigh its thrombotic risks. Whether the pattern of off-label use is similar in Canada, however, is not known. A focused review of the off-label use of rFVIIa in cardiac surgery at 18 Canadian hospitals found that its off-label use nearly doubled from 2003 to 2007. This increase was attributed to the late adoption at some centres for treatment of refractory hemorrhage.4 To determine the general pattern of off-label use in Canada since 2007, we conducted a comprehensive patient-level review of all off-label use of rFVIIa in non-hemophilia patients in surgical and non-surgical settings at 16 Canadian hospitals from 2007 to 2010. Additional objectives of this observational study were to explore the relationship between rFVIIa dose response and severity of bleeding and their relationships with major adverse events.

Methods

This was a historical cohort study. Investigators at 25 large Canadian hospitals were approached for participation; 16 agreed to participate and were included (Appendix). Institutional research ethics board approval was obtained at each participating hospital, all of which waived the need for informed consent. The blood bank or pharmacy database at each hospital was used to obtain a list of every patient who had received rFVIIa from January 1, 2007 to December 31, 2010.

Using a standardized data collection form, detailed data were retrospectively collected from medical records for all patients who had received rFVIIa while in hospital. Patients with hemophilia were excluded. Data were entered directly into a web-based database with built-in logic controls. Variables collected included patient demographics, comorbidities, hemodynamic and laboratory data, blood product transfusions (up to 24 hr before and after first dose of rFVIIa), indications for and dose of rFVIIa, as well as in-hospital (up to 30 days after therapy) major adverse events. Adverse events were obtained from patients’ clinical notes, discharge forms, death certificates, and autopsy reports where available. Major adverse events were divided into those likely to be thrombotic or non-thrombotic in nature. Stroke, deep vein thrombosis, pulmonary embolism, gut ischemia, limb ischemia, and myocardial infarction were classified into the thrombotic adverse events subgroup for analysis. Renal failure requiring dialysis, sepsis, disseminated intravascular coagulopathy, multi-organ failure, and cardiac arrest were classified into the non-thrombotic adverse events subgroup for analysis. Mortality was analyzed separately. The total number of red blood cells transfused within 24 hr of rFVIIa therapy was used as a surrogate measure for overall bleeding severity.

Statistical analysis

SAS™ version 9.3 (SAS Institute, Inc., Cary, NC, USA) was used for the statistical analysis. Categorical variables were summarized as frequencies and percentages and continuous variables as medians and interquartile ranges. As an indirect measure of effectiveness, the Wilcoxon signed-rank test was used to compare the amount of transfusions before and after rFVIIa therapy (data were collected for up to 24 hr before and after therapy). Only patients who underwent cardiac surgery were used in this analysis to provide for a more homogeneous sample of patients in whom rFVIIa was administered in response to postoperative bleeding.

The association between dose of rFVIIa (per kg) and severity of bleeding was measured with the Spearman’s rank correlation coefficient. For missing weight values, the gender-specific sample mean value was used. Dose of rFVIIa and severity of bleeding were categorized into quartiles, and their relationships with measured adverse events were assessed using the Mantel-Haenszel Chi square test. Multivariable logistic regression that controlled for severity of bleeding was used to determine the dose effect of rFVIIa on adverse events. The dose analyses were repeated after excluding patients with missing weight values, after excluding patients who received rFVIIa for non-surgical indications, and using the total dose rather than the weight-based dose.

Results

At the 16 participating hospitals, 1,378 patients received rFVIIa (median 87 patients, range 4-245 patients). Dosing data were missing in 11 patients, and these patients were excluded from the relevant analyses. The median [interquartile range] dose of rFVIIa (after imputing missing weight data in 106 patients) was 57 [36-85] µg·kg−1. Based on this distribution, patients were classified into quartiles as follows: < 36 µg·kg−1 (n = 352), 36-57 µg·kg−1 (n = 329), > 57 and < 85 µg·kg−1 (n = 352), and ≥ 85 µg·kg−1 (n = 334). Overall, 906 patients (66%) received a single dose of rFVIIa, 369 (27%) received two doses, 70 (5%) received three doses, 19 (1%) received four doses, and five (0.4%) received more than four doses. There was no discernible change in the amount or number of doses administered during the study period (data not shown).

The number of patients who received rFVIIa increased slightly from 2007 to 2008 but decreased thereafter (Fig. 1). Cardiac surgery was the indication for 72% of the cases (n = 987/1,378), increasing from 60% of the cases in 2007 (n = 201/341) to 81% in 2010 (n = 247/307) (Fig. 1). Of the 131 patients (10%) who received rFVIIa for non-surgical indications, 58 (44%) received it for intra-cerebral hemorrhage, 26 (20%) for reversal of warfarin, 24 (18%) for gastrointestinal bleed, and 12 (9%) for disseminated intravascular coagulation. Overall, the cohort consisted of high-risk patients (older, multiple comorbidities, high rate of catastrophic events, and requiring hemodynamic support prior to rFVIIa therapy) who had major blood loss (as measured by number of transfusions) and a high rate of adverse events (Table 1). There were no discernible changes in patient characteristics during the study period (data not shown). Coagulation tests conducted during the hour before rFVIIa therapy were, on average, not severely deranged, but test results were not available for many of the patients (Table 1). Comparing the amount of transfusions before and after rFVIIa therapy in patients who underwent cardiac surgery, we found that transfusions were substantially lower after rFVIIa therapy (Table 2).

Fig. 1
figure 1

Number of patients treated with rFVIIa per year rFVIIa = recombinant activated factor VII

Full size image
Table 1 Patient variables
Full size table
Table 2 Transfusions before and up to 24 hr after first dose of rFVIIa in the subgroup of patients who received rFVIIa due to bleeding after cardiac surgery
Full size table

The total dose of rFVIIa and bleeding severity in the entire cohort were correlated with each other (Spearman correlation coefficient 0.24; P < 0.0001), and adverse events generally increased in direct proportion to the increasing dose and bleeding severity (Figs. 2 and 3). After adjusting for bleeding severity, which remained directly related to adverse events, there was no dose-effect relationship between rFVIIa and any of the major adverse events (Fig. 4). The results were similar when patients with missing weight were excluded, when patients who received rFVIIa for non-surgical indications were excluded, or when total rather than weight-based dosing was used in the analyses (data not shown).

Fig. 2
figure 2

Relationship between total rFVIIa dose and adverse events Dose of rFVIIa quartiles are as follows: Quartile 1 is < 36 µg·kg−1 (n = 352); Quartile 2 is 36-57 µg·kg−1 (n = 329); Quartile 3 is > 57 and < 85 µg·kg−1 (n = 352); and Quartile 4 is ≥ 85 µg·kg−1 (n = 334). rFVIIa = recombinant activated factor VII

Full size image
Fig. 3
figure 3

Relationship between bleeding severity and adverse events Peri-rFVIIa red blood cell (RBC) transfusion includes all transfusions up to 24 hr before and after rFVIIa therapy: Quartile 1 is ≤ 6 units (n = 410); Quartile 2 is 7-10 units (n = 307); Quartile 3 is 11 to 17 units (n = 308); and Quartile 4 is > 17 units (n = 353). rFVIIa = recombinant activated factor VII

Full size image
Fig. 4
figure 4

Relationship between total rFVIIa dose and adverse events, adjusted for bleeding severity See Figs. 2 and 3 for quartile ranges. For both variables, Quartile 1 is the reference category. rFVIIa = recombinant activated factor VII; RBC = red blood cell

Full size image

Discussion

In this comprehensive four-year Canadian review, we found that the overall off-label use of rFVIIa in non-hemophilia patients remained relatively stable over the study period. We also found that rFVIIa was primarily and increasingly used in the setting of bleeding after cardiac surgery. The absolute use of rFVIIa for cardiac surgery increased to such an extent during the study period that it represented over 80% of usage during the last year of the study. We also found that major adverse events in the entire cohort were strongly related to the severity of bleeding rather than to the dose of rFVIIa.

One of the main concerns about rFVIIa are reports that its off-label use has been rapidly expanding since initial approval for treatment of bleeding episodes in hemophilia patients. This has led to calls for its manufacturer to be investigated for promoting the misuse of the drug.7 Our study, however, found that the overall off-label use of rFVIIa from 2007 to 2010 did not expand in Canada (although there was a change in pattern of usage with increased use in cardiac surgery). This finding is not consistent with the experience in United States hospitals where there was a 140-fold increase in the off-label use of rFVIIa from 2000 to 2008, with no indication of abatement in the latter years.3 While our study did not capture all rFVIIa use across Canada, the results are corroborated by the reduction in rFVIIa issued by the Canadian Blood Services (CBS; the supplier of the drug to all Canadian provinces except Quebec) from 2008 to 2011.8 As reported in the recent review by Lin et al.,8 the total amount of rFVIIa issued by the CBS reached a plateau of just over 32,000 mg during 2007 and 2008 and then gradually declined to about 27,000 mg in 2011. Assuming that the on-label use remained constant, this would imply that the off-label use decreased from 2008 to 2011. Our finding of reduced off-label use is not unique. A large (n = 3,314) registry that reported on all off-label use of rFVIIa at Australian and New Zealand hospitals from 2000 to 2009 also found a slight reduction in the use of the drug from 2006 to 2009.Footnote 2 It seems, therefore, that the concern about the “runaway” use of rFVIIa in off-label settings7 is not warranted, at least outside of the United States.

Another important finding of our study is that rFVIIa was primarily and increasingly used in the setting of severe bleeding in cardiac surgery where, in many cases, the bleeding was not responsive to standard therapy (as reflected by the relatively normal coagulation tests before rFVIIa therapy). Unlike most other off-label settings, cardiac surgery is an area for which the use of rFVIIa is supported by several observational studies and recommended in the setting of non-responsive bleeding.4,12,16-19 As in previous studies in cardiac surgery where the amount of transfusions before and after rFVIIa therapy was used as an indirect measure of effectiveness,4 we also found that the amount of transfusions in cardiac surgical patients was substantially lower after therapy than before therapy (Table 2). While this analysis is confounded by time (i.e., all bleeding eventually stops) and cannot prove causation, it is suggestive that rFVIIa may contribute to this bleeding cessation.

Another major concern with the off-label use of rFVIIa is its propensity to increase the risk of thrombotic adverse events in non-hemophilia patients without any clear benefit in reducing mortality. This has led to the recommendation that its off-label use be curtailed.8 It has also been argued, however, that the judicious off-label use of rFVIIa may be reasonable in patients who develop severe coagulopathic bleeding that is non-responsive to conventional therapies.10-15,18,19 The basis for this argument is primarily because rapid control of bleeding is crucial for reducing poor outcomes in this setting and because there is increasing evidence (albeit primarily from observational studies) that rFVIIa may achieve this objective.4,11 This latter argument is supported by this study’s finding that the rate of adverse events in the entire cohort was strongly related to bleeding severity, to such an extent that the relationship dwarfed any potential risks attributable to the dose of rFVIIa. Further in support of this argument, the Australian and New Zealand registry on the off-label use of rFVIIa found a clear relationship between lack of response to rFVIIa and mortality. Specifically, analyzing data from over 2,500 patients, they found that 62% (n = 479/771) of patients who did not respond to rFVIIa died, whereas only 20% (n = 363/1,818) of those who did respond died (P < 0.001).A

Our study has several important limitations. First, a large number of data are missing for some variables, and there is the potential for inaccurate or incomplete data collection. To ensure that serious adverse events were accurately captured, we limited our data collection to adverse events that were expected to be recorded accurately in patients’ records. Moreover, we had to use surrogate measures for some important variables, such as severity of bleeding, and could not capture some other important variables, such as the reasons for dose selection. Second, as this was a historical cohort study, causation cannot be proven. Third, as this was a registry, we had access only to data on patients who received rFVIIa. Lacking a control group, the study was not equipped to determine the association of rFVIIa therapy itself with outcomes. Thus, only the dose-response relationship between rFVIIa and measured outcomes could be explored. Fourth, the study was sponsored by the makers of rFVIIa, which may have introduced bias; however, the support was through an unrestricted grant and the sponsor had no input into the conduct of the study, the analysis of the results, or the writing of the manuscript.

In conclusion, this registry showed that the off-label use of rFVIIa in Canada remained relatively stable over a four-year period from 2007 to 2010 (although there was an absolute increase in its use in cardiac surgery). We also found that major adverse events in patients receiving rFVIIa were related to the severity of bleeding and not to the dose of rFVIIa. These findings add credence to the argument that use of potent hemostatic agents, such as rFVIIa, to gain rapid control of bleeding that is non-responsive to conventional therapies may at times be warranted.