Response to treatment and adverse events associated with use of recombinant activated factor VII in children: a retrospective cohort study

Abstract

Background:

Recombinant activated factor VII (rFVIIa) is United States (US) Food and Drug Administration (FDA)-approved for patients with hemophilia with inhibitors or congenital factor VII deficiency. Initial reports of off-label use highlighted its efficacy, though newer reports have not repeated these findings. In both types of publication, though, secondary thromboses have been seen in adult patients. The data in children are less clear.

Methods:

This study analyzed all rFVIIa use at a large children’s hospital for characteristics and outcomes. Recipients of rFVIIa were identified retrospectively via the electronic medical record. Data on patient diagnosis, lab data, other treatments, adverse events, and outcomes were collected.

Results:

Over 33 months, 66 patient episodes were treated with a total of 606 doses (median = 2). The most common indication (36.4%) was gastrointestinal bleeding (24/66 patients). Only one patient received a dose for an approved labeled indication. For control of bleeding, 33.3% of courses were unsuccessful (19/57). Bleeding from multiple sites was associated with treatment failure. In 16.7% of patients (11/66), unexpected adverse thromboses developed within 1 week of completing a course of rFVIIa. Thromboses in both intra- and extra-corporeal sites were included if they compromised patient care.

Conclusions:

In the majority of cases reviewed, rFVIIa was successful in stopping or slowing serious bleeding episodes. It was least effective when a patient had diffuse bleeding at the time of administration. The thrombosis rate of 16.7% was higher than expected, though causality cannot be declared. Further investigation is needed to determine the risk–benefit ratio in children.

Introduction

Recombinant activated factor VII (rFVIIa) (NovoSeven®, NovoNordisk) was developed and licensed in the United States (US) to treat and provide prophylaxis against bleeding in hemophilia with inhibitors and, subsequently, congenital factor VII deficiency. Recommended dosages are 90 µg/kg for hemophilia patients and 15–30 µ/kg for factor VII deficient patients, though minimum effective doses have not been established [Novonordiska, 2006]. In these settings, rFVIIa has been found to be efficacious and effective for treatment and prophylaxis of bleeding [Shapiro, 2000; Kavakli et al. 2006; Mathew and Young, 2006; Santagostino et al. 2006]. Later reports, however, have questioned this effectiveness [Yank et al. 2011; Logan et al. 2011].

Reports of off-label use of rFVIIa in pediatric patients have lagged behind those of adult patients, but case series of various sizes are emerging [Young et al. 2005; Chuansumrit et al. 2007; Reiter et al. 2007; Herbertson and Kenet, 2008; Okonta et al. 2012]. Most of the early reports documented favorable success rates in controlling nonhemophilia bleeding with rFVIIa. Again, newer pediatric studies report conflicting results [Guzzetta et al. 2012]. Conversely, the reported rates of adverse events in the pediatric literature have been very small. Reasons for these differences may relate to the shorter half-life of rFVIIa and its more rapid clearance from the circulation in pediatric patients [Erhardtsen, 2000]. On the other hand, the low early rate of reported thromboses may be a result of publication bias. Newer reports have shown an increased awareness of thrombotic complications [Witmer et al. 2011; Mcquitten et al. 2012].

With this study, we sought to contribute to the emerging data on off-label rFVIIa use in children by analyzing all rFVIIa use at one institution. The primary objective was to assess the differences between patients who responded and those who did not respond when rFVIIa was given for off-label indications. Secondarily, we sought to investigate clinical and laboratory data about pediatric patients who did and did not develop adverse events.

Methods

The Institutional Review Board at the University of Pittsburgh, PA, USA approved this study as an exempt protocol. Individual consent was not required due to the de-identification of all patient data; the Department of Pharmacy reported a list of all patients who received rFVIIa between January 2005 and September 2007. Medical records from 7 days prior to the first dose through 7 days after the last dose were reviewed on all patients. No patients were excluded from the analysis. Sample size and power calculations were not performed prior to data collection.

Data collection

Using a standard evaluation form, a broad panel of data was recorded for every episode of rFVIIa use. Independent variables based on clinical data included patient age, infancy status (defined as age <1 year), sex, weight, admitting diagnoses/services, bleeding indication(s), dose of rFVIIa, dosing interval, and transfusion history [fresh frozen plasma (FFP), cryoprecipitate, and packed red blood cells (PRBCs)] for the 7 days preceding the first rFVIIa dose. Bleeding indications were then reclassified into a categorical variable to show whether patients had multiple or a single indication for receiving rFVIIa. Variables based on laboratory data included prothrombin time (PT), partial thromboplastin time (PTT), fibrinogen, white blood cell count (WBC), hemoglobin, platelet count, alanine aminotransferase (ALT), total bilirubin, blood urea nitrogen (BUN), creatinine, and blood pH (arterial or venous). The latest values preceding the first dose of rFVIIa were recorded as baseline.

Whether a patient received rFVIIa for the treatment of active bleeding or for prophylaxis against bleeding was determined by a review of notes in the medical record on the day prior to and the day of medication administration. Subsequently, three dependent variables were assessed from the clinical record: response to rFVIIa (as determined by bleeding cessation), development of a venous or arterial thrombosis at any point from the administration of the first dose to within 7 days of the final dose of rFVIIa, and mortality within 7 days of the final dose. The primary outcome was bleeding cessation, while the two adverse event outcomes were secondary.

As there are no objective methods to assess therapeutic response to rFVIIa, the primary author classified this outcome variable into two categories: response to treatment (grouping full and partial bleeding cessation) or no response to treatment. This assessment was based upon notes in the medical record by nurses, residents, fellows, and attending physicians. Mortality was objectively defined using death summaries or autopsy reports in the medical record. The primary investigator reviewed radiographic or autopsy reports in the medical record to confirm the presence or absence of a thrombosis.

Data analysis

Descriptive statistics were calculated for all variables; however, only the 57 episodes that involved treatment of bleeding were included in the analysis of the primary outcome. The data were assessed for statistically significant differences between the responders and nonresponders, those with and without thromboses, and those who survived and those who died. Mann–Whitney tests were performed to compare median values of all continuous variables. Chi-square tests were used to analyze the following categorical variables: age <1 year (therapeutic response), male sex (therapeutic response), transplant surgery patients (all outcomes), presence of gastrointestinal (GI) bleeding (therapeutic response and mortality), presence of multiple sites of bleeding (therapeutic response), and pre-rFVIIa transfusions of blood products (therapeutic response and mortality). Fisher’s exact tests were performed for all remaining comparisons. A multivariable logistic regression was performed when the number of events allowed adequate model analysis. An alpha level of 0.05 was used to determine significance for all tests.

Statistical packages used for the analysis included Stata 10 (Stata Corporation, College Station, TX, USA) and SPSS 16 (SPSS Inc., Chicago, IL, USA).

Results

Over 33 months, there were 66 patient care episodes that used a total of 606 doses of rFVIIa. In 9 episodes, rFVIIa was used as a prophylaxis against bleeding. The remaining 57 episodes were treatments to control active bleeding. Overall, only one use of rFVIIa in this study period was for a US Food and Drug Administration (FDA)-approved indication: prophylaxis for a child with mild congenital factor VII deficiency.

Patient and episode characteristics

The median age of patients who received rFVIIa was approximately 2 years (range 1 day to 19.5 years). A total of one-third of all patients were less than 1 year of age (22/66). Males comprised nearly two-thirds of all patients in the cohort. A total of 30 children out of 66 (45.5%) were abdominal viscera transplant candidates or recipients (liver, small bowel, or multivisceral). They comprised the largest group of patients to receive rFVIIa. The second largest group of patients had oncologic diagnoses (Figure 1). Over 75% (50/66) of all patients received transfusions with PRBCs, platelets, and FFP before being treated with rFVIIa. However, fewer than half of patients received transfusions of cryoprecipitate.

Figure 1.

Admitting diagnoses by disease service for the patients who received rFVIIa.

Overall, 57 episodes occurred in the Pediatric Intensive Care Unit, 3 on the general medical floors, and 2 each in the operating room, Neonatal Intensive Care Unit, and Cardiac Intensive Care Unit. The most common indication was GI hemorrhage (Figure 2). Over 25% (17/66) of patients had multiple indications or bleeding at multiple sites (i.e. GI bleeding and pulmonary hemorrhage or hematuria and intracranial hemorrhage) at the time of their first dose of rFVIIa. Only 4 of the 66 episodes were given for bleeding sustained in trauma. Only one patient received a dose for a labeled indication (congenital factor VII deficiency); no patients with hemophilia and inhibitors received doses during the study period.

Figure 2.

Frequencies of indications for use of rFVIIa. More than one indication was present in certain patients. A total of 22 different bleeding indications were described in the medical record across the group of 66 patients and 17 patients were listed as having ⩾2.

Dosing information

The mean and median doses of the entire cohort were 83.1 and 87.1 µg/kg respectively (range 21.1–124.7 µg/kg). The number of doses administered to patients ranged from 1–194, with a median of 2 and a mean of 9.18. A total of 67% of patients received three or fewer doses. The patient who received 194 doses was a marked outlier in the dataset (average dose was 101.8 µg/kg). This patient had a failing intestinal graft and chronic mucosal bleeding that was felt to be refractory to all other management strategies by the primary service. If this patient is excluded from the dataset, the mean and median doses are changed slightly to 82.8 and 89.3 µg/kg (range was unchanged). The median and mean number of doses when the outlier is excluded are reduced to 2 and 6.33 (range, 1–83).

Primary outcome

Among the 57 patients treated for active bleeding, 33% (19 total) did not have any response to treatment with rFVIIa. There were four statistically significant differences found among the independent variables. First, a much higher proportion of patients who did not respond had bleeding from multiple locations, 52.6% (10/19) versus 18.4% (7/38); p = 0.01. Second, a much higher proportion of responders were male, 76.3% (29/38) versus 47.4% (9/19); p = 0.03. Third, patients who responded to treatment were more likely to be <1 year of age than those who did not respond, 42.1% (16/38) versus 15.8% (6/19); p = 0.05. Finally, patients who responded to treatment had a lower median weight than those that did not respond (10 kg versus 18 kg;p = 0.04). There were no other statistically significant differences between the two groups. Multivariable analysis showed that female sex and bleeding from multiple locations were significantly associated with lack of clinical response (p = 0.01). No other variables in multivariable regression had significant results.

Adverse events

Overall, 11 of the 66 patients (16.7%) developed an unexpected adverse thrombosis within 7 days of completing their rFVIIa course (Table 1). A total of two of these occurred in patients who received rFVIIa for prophylactic use only. When rFVIIa was used for treatment of bleeding, thromboses appeared in 15.8% among responders and nonresponders (6/38 versus 3/19). A total of two patients were found to have thromboses in a visceral space; two others had the thromboses in an extracorporeal circuit or devices. These cases were included in the definition of unexpected adverse thrombosis because their presence complicated the care of each affected patient.

Table 1.

Characteristics of patients who developed thromboses associated with rFVIIa use.

Patient	Age, years	Indication for rFVIIa use	Clot location	Timing/clot found	Survival at 7 days
1	16.5	Sinus of Valsalva rupture	Mediastinum after cardiac surgery	Within 48 h	Alive
2	13.33	Post-operative chest tube bleeding	Pleura and mediastinum after cardiac surgery	Within 12 h	Alive
3	6.67	GI bleeding	SMV, common femoral vein, superficial femoral vein	Within 7 days	Alive
4	4.75	Prophylaxis against bleeding	External iliac vein (CVL-associated)	Within 24 h	Alive
5	15.5	GI bleeding	CVVH circuit	Within 24 h	Alive
6	3.08	Cerebral hemorrhage	EVD	Within 1 h (during surgery)	Dead, other causes
7	14.33	Hematuria & intra-abdominal bleeding	LLL segmental pulmonary embolus	Within 72 h	Alive
8	10.42	GI bleeding	Renal artery	Within 48 h	Alive
9	2	Prophylaxis against bleeding	Hepatic artery	Within 5 days	Alive
10	1	Tracheostomy bleeding	Small vessel pulmonary emboli	Autopsy (died within 2 days)	Dead, other causes
11	0.42	GI bleeding	Internal jugular vein (CVL-associated)	Within 7 days	Alive

CVL, central venous line; CVVH, continuous veno-venous hemofiltration; EVD, extraventricular drain; GI, gastrointestinal; LLL, left lower lobe; rFVIIa, recombinant activated factor VII; SMV, superior mesenteric vein.

Median patient weight was found to be significantly higher among those who developed thromboses (10.4 kg versus 23.4 kg; p = 0.03); however, median dose per weight was significantly lower in this group (82.8 µg/kg versus 89.7 µg/kg; p = 0.03). Comparison of baseline laboratory values found that the PTT (30.7 s versus 43.9 s; p = 0.002) and the ALT (47.5 IU/l versus 104 IU/l;p = 0.03) were both significantly lower in the group that developed thromboses (Table 2).

Table 2.

Baseline laboratory characteristics of patients who developed thromboses after receiving rFVIIa.

Patient	Doses(#)	Dose per weight (µg/kg)	PT(sec)	PTT(sec)	WBC (10A9/l)	Hemoglobin (9/dl)	Plateletcount (10A9/l)	Fibrinogen (mg/dl)	BUN (mg/dl)	Creatinine (mg/dl)	ALT(IU/l)	Total bilirubin(mg/dl)
1	2	82.76	13.5	19	31.7	11.1	270	166	12	1.1	31
2	2	21.13	10.2	31.1	10.8	11.9	72	227	27	1.3	29
3	194	101.18	12.6	24.7	19.1	9.9	161	234	9	0.4	83	2
4	1	68.97	16.9	30.7	7.3	9.8	83	173	7	0.3	80
5	2	41.74	13	28	0.1	9.3	42	912	105	2.9
6	2	78.33	8.3	28.3	3.3	7.8	44	223	10	0.3	17
7	2	85.71	10	29.8	8.9	20.6	81	511	51	0.3	27	0.8
8	1	82.29	11.3	31.1	2.2	15.3	92	227	52	1.2	29	5.7
9	1	83.21	18.4	37.9	3	8.9	57	209	16	0.5	74	18.6
10	1	86.3	31.9	120.9	9.3	13.3	62	289	14	0.4	336	23
11	6	84.38	13	39.1	5.9	8	52		36	0.2	56	4.8

ALT, alanine aminotransferase; BUN, blood urea nitrogen; PT, prothrombin time; PTT, partial thromboplastin time; rFVIIa, recombinant activated factor VII; WBC, white blood count.

While none of the thromboses could be definitively attributed directly to rFVIIa, review of clinical documentation indicated that they all were associated with rFVIIa use (Table 1). Only two patients with thromboses died, though the deaths were attributed to other causes at autopsy.

Overall cohort survival within 7 days of the last dose of rFVIIa was 78.8% (52/66). Among those who responded to treatment, survival was 86.8% (33/38) while in those who did not respond, it was 57.9% (11/19), p = 0.01. Other than treatment response, no other clinical or laboratory value was found to be significantly associated with survival.

Discussion

During recent military conflicts in the Middle East, both the US and Israeli Armies began to use rFVIIa to treat hemorrhage sustained from combat trauma. In 2004, the US Army declared rFVIIa a standard of care in treating battlefield hemorrhage based upon unpublished data from international trials [Little, 2006]. This helped create a reputation for rFVIIa as a ‘universal hemostatic agent’, leading to skyrocketing use in patients with bleeding due to other etiologies. In 2005, it was reported that over 90% of rFVIIa use in the world was for non-FDA-approved indications [Maclaren et al. 2005]. The literature soon became populated with positive case series and small clinical projects extolling the virtues of rFVIIa for a multitude of bleeding scenarios. However, subsequent publications have provided contradictory results, often leading to controversies within institutions about this drug’s merits [Logan et al. 2011; Witmer et al. 2011; Yank et al. 2011; Guzzetta et al. 2012; McQuitten et al. 2012; Okonta et al. 2012].

At the same time as providers debate its true efficacy, rFVIIa has been found to carry a moderate amount of risk. In the same 2005 study that summarized off-label use, it was reported that 9.8% of rFVIIa courses resulted in adverse events, most of which were thromboembolic [Shander et al. 2005]. Similarly, a study of rFVIIa for intracerebral hemorrhage also found a 10% incidence of thromboses, despite an overall favorable conclusion [Brown and Morgenstern, 2005]. In December 2005, the US FDA issued an alert to warn adult patients about the risk of myocardial infarction and stroke when receiving rFVIIa in off-label settings. These concerns were formalized in a January 2006 study that detailed the time-dependent risk of thromboses when adult patients received rFVIIa for nonhemophilia indications [O’Connell et al. 2006].

Our study reports a cohort of pediatric patients who received rFVIIa for various diagnoses, the vast majority of which were not US FDA-approved indications. Only one patient received a dose for a labeled indication (congenital factor VII deficiency); no patients with hemophilia and inhibitors received doses during the study period. The lack of hemophilia patients in the study period is likely due to chance and a reflection of the 33-month time frame (our institution also cares for a high percentage of patients with mild or moderate factor IX deficiency, who have lower rates of inhibitor development than patients with factor VIII deficiency). Initially, the literature for off-label use in pediatric patients consisted mainly of case reports and small series that suggested good efficacy. Reviews of these studies have been published elsewhere [Mathew et al. 2003; Mathew, 2004; Young et al. 2005]. In larger series, many children received a favorable response but 6–22% did not see any benefit [Young et al. 2005; Chuansumrit et al. 2007; Reiter et al. 2007; Heller et al. 2008]. Newer, even larger, reports suggest the early reports on the benefits of rFVIIa have been overstated [Witmer et al. 2011; Yank et al. 2011; McQuitten et al. 2012].

We found a higher than expected treatment failure rate plus the highest rate of rFVIIa-associated thromboses reported to date. Our study addresses the concern of publication bias and provides evidence for several new conclusions about off-label rFVIIa use in children. In our study, 16.7% of children developed an unexpected adverse thrombosis within 1 week of receiving their final dose of rFVIIa (across all indications for its use). This rate is higher than estimates from adult populations and in previous pediatric reports. Thromboses occurred in both venous and arterial circulations, in surgical spaces, and extracorporeal circuits. All of these complicated patient care in some manner, though none of the thromboses were fatal; all thromboses were included in the calculation of the overall rate because they were felt to represent a direct adverse effect of rFVIIa at the time of patient care. Of concern, however, is that they were not restricted to vessels with central venous access devices, the most common risk factor for thromboses in children. This suggests that rFVIIa has far-reaching coagulant activity and does not act at only those sites with known vascular damage.

Conversely, none of the baseline laboratory parameters we surveyed were found to be significantly different between children who responded to treatment and those who did not. It was seen, however, that responders were more likely to be male, be >1 year of age, have lower weights, and have bleeding from only one location. The significance of male sex is unclear. While there are known sex differences in the coagulation system, the laboratory data in our study revealed only one. Males had a lower fibrinogen level than females (p = 0.05). However, median levels were within the normal range (177.5 mg/dl versus 240 mg/dl). Unfortunately, we are unable to explain the sex difference noted in our study.

The lower average weight among responders is likely to be a reflection of the better response rate of infants when compared with children. Taking response to treatment out of consideration, patients <1 year were found to have a significantly higher PTT and hemoglobin, though lower serum creatinine. The hemoglobin may reflect disease-specific changes, though the variation in creatinine and PTT may be related to age-dependent changes. When the response to treatment was examined separately in each age category, there were no significant differences that could be viewed as an explanation for the better response in infants. Consequently, the significance of this finding also remains unclear.

On the other hand, the presence of bleeding from multiple sites and its relation to treatment response is a potentially important finding. Over half of the patients who failed to respond had bleeding from multiple sites. The odds of responding to rFVIIa when there were multiple areas of bleeding were approximately one-fifth the odds of responding when there was only one site of bleeding.

There have been reports discussing futility and early versus late use in off-label settings [Clark et al. 2004; Perkins et al. 2007]. These studies have suggested that there may be a point during a hemorrhagic event beyond which rFVIIa cannot be expected to show benefit. The etiology of hemorrhage often is not known when rFVIIa is used in an off-label fashion. While it is understood that rFVIIa gains its effect from a thrombin burst generated from the interaction with an activated platelet surface, the underlying pathophysiology of children with diffuse hemorrhages is not always delineated. Our data that rFVIIa is not as effective in children with widespread bleeding support the belief that this medication should not be used in futile situations.

Our study does have several limitations due to its nature as a retrospective study. First, patients did not receive the same dose, schedule, or pre-rFVIIa transfusions. The attribution of rFVIIa as the reason that bleeding was controlled was based on an interpretation of the medical record, not direct observation. Secondly, the same laboratory panels were not performed on all patients pre- and post-rFVIIa, though there are no evidence-based guidelines to suggest any laboratory studies are required before its use. The discrepancy in laboratory evaluation precludes a more accurate comparison of each patient’s baseline coagulation status. Thirdly, our results cannot be used to declare causation. The thromboses can only be described as associations. Fourth, the number of patients in certain groups was relatively small. One could question that a larger population of patients could have produced a greater number of predictive factors. A larger population, with more standardized observations, could allow for more powerful statistical evaluation, such as a multivariable analysis, which our study was not set up to perform.

Another potential limitation comes from our ending the observation period for thrombosis monitoring 7 days after the final dose (7 days was chosen due to the short half-life of rFVIIa). Additionally, it was felt that with the majority of patients located in critical care units, the number of potentially confounding contributing causes to thrombosis development would increase if the observation period was extended. A prospective and more controlled study may have detected even more thromboses.

Our population also was different from many of those reported in previous studies. We had very few trauma patients (only four) and a large group of children on the transplant surgery service. However, the majority of those patients were transplant candidates, who frequently develop acute illnesses at smaller hospitals prior to transfer to a larger institution. In that sense, our findings may be generalizable to select patients at nontertiary care hospitals. Our results also include one patient who received a staggering 194 doses. This case, in reviewing the detailed data, featured intractable mucosal bleeding in a failing intestinal graft. The case also featured significant debate among the various specialties caring for the patient in determining the best course of management. Though the number of doses is very high, we elected to include this patient in the data analysis because the underlying diagnosis is characteristic of our institution and shows the temptation and frustrations of using a celebrated, at the time, medication for an unlabeled indication. A higher than average dose of rFVIIa also was administered to this patient. Unlike the decisions to administer such frequent doses, the average dose was a reflection of vial sizes available at that time. The fact that both the dose and number of doses are highest in the same patient is likely a coincidence.

Conclusion

This study reports a series of children who received rFVIIa and had a lower than expected response rate and a higher than expected rate of thromboses. The ability to treat critically ill patients with severe bleeding would benefit from future prospective studies. A better physiologic understanding of diffuse hemorrhage as well as a better understanding of the precise mechanism of action for rFVIIa in nonreplacement or inhibitor-bypassing situations would contribute significantly to better outcomes.