Article, Hematology

Risk of thromboembolic events after protocolized warfarin reversal with 3-factor PCC and factor VIIa

a b s t r a c t

bleeding events and life-threatening hemorrhage are the most feared complications of warfarin therapy. Prompt Anticoagulant reversal aimed at replacement of vitamin K-dependent clotting factors is essential to promote hemostasis. A retrospective cohort study of warfarin-treated patients experiencing a life-threatening hemor- rhage treated with an institution-specific warfarin reversal protocol (postimplementation group) and those who received the prior standard of care (preimplementation group) was performed. The reversal protocol included vitamin K, 3-factor prothrombin complex concentrate, and recombinant factor VIIa. Demographic and clinical information, anticoagulant reversal information, and all adverse events attributed to warfarin reversal were recorded. A total of 227 patients were included in final analysis, 109 in the preimplementation group and 118 in the postimplementation group. Baseline patient characteristics were similar in both groups, with the ex- ception of higher average Sequential Organ Failure Assessment scores in the postimplementation group (P =

.0005). The most common indication for anticoagulation reversal was intraparenchymal hemorrhage. Prereversal international normalized ratios (INRs) were similar in both groups. Attainment of INR normalization to less than

1.4 was higher, and rebound INR was lower in the postimplementation group (P b .0001; P = .0013). Thrombo- embolic complications were significantly higher in the postimplementation group (P = .003). Elevated baseline Sequential Organ Failure Assessment score and mechanical valve as an indication for anticoagulation were inde- pendently associated with thrombotic complications (P = .005). A warfarin reversal protocol consisting of 3- factor prothrombin complex concentrate, recombinant factor VIIa, and vitamin K more consistently normalized INR values to less than 1.4 as compared to the prior standard of care in a diverse patient population. This success came at the cost of a 2-fold increase in risk of Thromboembolic complications.

Introduction

Warfarin, the predominant oral anticoagulant prescribed in the United States for the past half century, is indicated for the treatment and prevention of thromboembolic complications [1]. Although warfa- rin use has begun to decline with the recent introduction of novel oral anticoagulants, it still remains one of the most highly prescribed medi- cations in the United States, totaling greater than 30 million prescrip- tions annually [2-4].

? Conflicts of interest and financial support: None.

?? This article has not been presented in any other format.

* Corresponding author at: Department of Pharmacy, Oregon Health & Science Univer- sity, 3181 SW Sam Jackson Park Rd, CR 9-4, Portland, OR 97239. Tel.: +1 503 494 4660.

E-mail addresses: [email protected] (C.A. Barton), [email protected] (N.B. Johnson), [email protected] (J. Case), [email protected] (B. Warden), [email protected] (D. Hughes), [email protected] (J. Zimmerman), [email protected] (G. Roberti), [email protected] (W.D. McMillian), [email protected] (M. Schreiber).

Warfarin-associated bleeding events occur in 15% to 20% of patients, with life-threatening hemorrhage occurring in 1.7% to 3.4% of patients annually [5-7]. Patients who present with acute hemorrhage require prompt anticoagulant reversal via replacement of vitamin K-dependent coagulation factors. Several pharmacologic and blood product options exist to replace these factors, which include vitamin K, recombinant factor VIIa (rFVIIa), fresh frozen plasma , and prothrombin complex concentrates [8].

Vitamin K promotes hepatic production of clotting factors II, VII, IX, and X. Supplementation is necessary to reverse warfarin-related anticoagulation and provide a sustained effect [1,9]. The production of vitamin K-dependent clotting factors is delayed after vitamin K supple- mentation, and this necessitates administration of exogenous clotting factors, via FFP or PCC, for immediate reversal in the setting of life- threatening hemorrhage [1].

Fresh frozen plasma is a human plasma product containing homeo- static concentrations of clotting factors [1,9]. There are several

http://dx.doi.org/10.1016/j.ajem.2015.06.010

disadvantages of FFP including large infusion volumes, delays in Time to administration due to ABO typing and product thawing, infection, transfusion-related acute lung injury, and transfusion-associated cardi- ac overload [1,9].

The major alternatives to FFP are PCCs. Prothrombin complex con- centrates used for anticoagulation reversal are plasma-derived, concen- trated mixtures of vitamin K-dependent clotting factors. Three-factor PCCs (3F-PCC) contain significant quantities of nonactivated factors II, IX, and X and minimal amounts of factor VII [1].

Our institution developed a standard treatment algorithm for anticoagulation reversal using 3F-PCC (Profilnine^(R) SD, Grifols, Los Angeles, CA), fixed-dose rFVIIa, and vitamin K. Patients receiving an oral anticoagulant before admission who presented with life- threatening bleeding and an International normalized ratio great- er than or equal to 1.5 were eligible for reversal with this protocol. The protocol consisted of 50 U/kg 3F-PCC (Profilnine^(R) SD; maximum, 4000 U), 1-mg fixed-dose rFVIIa, and 10 mg of intravenous vitamin K. A sec- ond dose of 25 U/kg PCC was available should the initial reversal not achieve an INR of less than or equal to 1.4. This protocol and others closely resembling it have been reported to be effective in the reversal of anticoagulation in warfarin-related Intracranial Hemorrhage [10-14]. The objective of this study was to evaluate the safety and effica- cy of this protocol used in a diverse patient population as compared to the prior standard of practice for warfarin-associated bleeding reversal.

Methods

An institutional review board-approved, retrospective cohort study of all patients with life-threatening warfarin-associated bleeding epi- sodes admitted from March 1, 2008, to February 28, 2014, was complet- ed. Patients were included if they were 18 years or older and required warfarin reversal with a factor-containing product. Patients were ex- cluded from the study if they received anticoagulation reversal for any indication other than active bleeding, died within 12 hours of admis- sion, or the INR was deemed fully reversed before transfer to Oregon Health & Science University.

Patients treated before implementation of the reversal protocol (“preimplementation” group) were admitted between March 2008 and February 2011 and identified via International Classification of Dis- eases, Ninth Revision (ICD-9), coding for life-threatening bleeds. Search terms for ICD-9 coding included hemorrhage, hemorrhagic, and bleed. Patients treated with the reversal protocol (“postimplementation” group) were admitted between March 2011 and February 2014 and identified via a search of the electronic medical record for all patients older than 18 years who were administered Profilnine SD or rFVIIa.

Subject inclusion and exclusion criteria were applied, and all data were collected via extensive medical record review. Data collected for all patients included demographic data; clinical, radiologic, and labora- tory data; coagulation values; length of stay in hospital and intensive care unit; type and location of hemorrhage; surgical information; rever- sal agent and blood product administration; occurrence of rebound INR; indication for warfarin use; antiplatelet use; and any complications at- tributed to reversal agent or blood product administration.

clinical efficacy was assessed based on the type of bleeding event. For Intracranial bleeding events, Hematoma expansion on repeated im- aging was evaluated; for gastrointestinal and traumatic bleeding events, packed red blood cell (pRBC) Transfusion requirements before and after reversal were calculated. In the preimplementation group, all pRBC use within the first 12 hours after the bleeding event were considered to be “prereversal” requirement, and the postreversal requirement was re- corded for 36 hours after reversal. In the postimplementation group, all pRBC use was recorded before reversal, and postreversal require- ments were recorded for 48 hours after the bleeding event.

Rebound INR was defined as having achieved an INR les than 1.4 and then having a repeat INR within the next 48 hours greater than or equal to 1.4. Patients who did not achieve an INR of less than 1.4 after reversal

agents were administered were excluded from rebound INR analysis. In the preimplementation group, the number of units of FFP given for anticoagulation reversal before the first repeat INR value was obtained as well as the total number of units received within the first 24 hours of admission or first 24 hours after the bleeding event.

Complications were identified via extensive chart review of daily progress notes, discharge summary notes, radiographic findings, and laboratory values. specific complications included, but were not limited to, any form of venous thromboembolism, ischemic stroke, myocardial infarction, exacerbation of heart failure, pulmonary edema, rash, or ana- phylaxis. Screening for thromboembolic complications was performed at the discretion of the treating providers. All known Thromboembolic events that occurred after receipt of the reversal protocol were docu- mented. Only complications that occurred during the principal admis- sion were considered.

Data were analyzed using JMP 11.0.0 (Copyright 2013; SAS Institute, Inc, Cary, NC). Descriptive statistics were used to summarize patient de- mographics and outcomes. Data are presented as mean and SD or medi- an and interquartile range or percentages as appropriate. Continuous variables were tested for normality by use of the Shapiro-Wilk W test. Normally distributed continuous data were analyzed by the Student t test, whereas non-normally distributed continuous data were analyzed by the Wilcoxon rank sum test. ?² or Fisher exact test was used to com- pare nominal data as appropriate. An a priori ? level of less than or equal to .05 was used to determine statistical significance for all comparisons. To perform a multivariate logistic fit model for thrombotic complica- tions, patient characteristics found to be different at baseline with a predefined P <= .2 were entered.

Results

The ICD-9 coding for any type of hemorrhage identified 1976 pa- tients admitted between March 2008 and February 2011. We excluded 1795 patients not receiving Oral anticoagulation, 47 patients not actively reversed with blood or factor product, 16 patients who died within 12 hours of admission, 4 patients with incorrect ICD-9 codes who never ex- perienced an acute bleeding event, 3 patients who entered into our elec- tronic medical record but were never transferred from the referring facility, and 2 patients who transferred to our facility after complete re- versal and resolution of bleeding. A total of 109 patients were included for final analysis in the preimplementation group.

A search of all Medication orders for Profilnine SD^(R) and rFVIIa be- tween March 2011 and February 2014 identified 374 patients. The rever- sal protocol was not administered to 145 patients. A total of 111 patients were excluded: 52 patients were not receiving oral anticoagulation, 34 patients were treated for coagulopathy in the setting of traumatic injury, 11 patients died within 12 hours of admission, 6 patients were receiving novel oral anticoagulants, 4 patients were reversed for procedures, 2 pa- tients were reversed for parenteral anticoagulation, and 2 patients only received the rFVIIa portion of the reversal protocol. A total of 118 patients were included for final analysis in the postimplementation group.

Patient characteristics prereversal and postreversal protocol imple- mentation are displayed in Table 1. The average Sequential Organ Failure Assessment score was higher in the postimplementation group (Table 1). The most common indication for anticoagulation in both groups was atrial fibrillation, followed by venous thromboembolism and presence of mechanical heart valve (Table 1). The most common indica- tion for reversal in both groups was ICH, followed by gastrointestinal (GI) hemorrhage and acute traumatic hemorrhage. There were signifi- cantly more GI hemorrhages in the preimplementation group (P b

.0001) (Table 2). The most common type of ICH in both groups was intraparenchymal hemorrhage, followed by subdural hematoma and sub- arachnoid hemorrhage.

In the preimplementation group, most patients received at least 1 U of FFP (97.2%). An average of 2.97 U of FFP was given during the Initial resuscitation period, with an average of 3.86 U given in the first 24

Table 1

Baseline demographics

Table 3

Reversal agents administered and INR outcomes

	Preimplementation	Postimplementation	P			Preimplementation	Postimplementation	P
	(n = 109)	(n = 118)				(n = 109)	(n = 118)
Age, median (IQR)	77 (66-82)	73 (62-82)	.32		Reversal agent
Male sex, n (%)	61 (56.0)	71 (60.2)	.59		Vitamin K
Weight (kg), median (IQR)	80.2 (69.1-95.9)	81.7 (70.1-102.0)	.49 Any dose, n (%) 100 (91.7) 111 (94.0) .61
SOFA score, median (IQR)	2 (1-4)	4 (1-7)	.0005 FFP
GCS score, median (IQR)	15 (13-15)	14 (10-15)	.08	Any dose, n (%)		106 (97.2)	56 (47.4)
Hemoglobin level (g/dL),	10.6 +- 2.4	10.3 +- 2.3	.29	Mean dose within 24 h (U)		3.9	3.3
mean (SD)					rFVIIa
Hematocrit (%), mean (SD)	31.1 +- 7.2	30.6 +- 6.9	.56	Any dose, n (%)		28 (25.6)	111 (94.1)
Indication for anticoagulation, n (%)			.02	Mean dose (mg)		3.9	1
Prophylaxis 83 (76.1)		80 (67.8)	PCC

Treatment

26 (23.9)	34 (28.8)^a		Patients, n (%)	116 (98.3)
30 (27.5)	41 (34.7)	.06	Mean dose (U/kg)^a	3748 (44.7)
23 (85.2)	39 (95.1)		INR outcome

PTA antiplatelet agent, n (%) Aspirin 81 mg

Aspirin 162-325 mg	4 (14.8)	2 (4.9)	Prereversal INR, median (IQR)	3.0 (2.2-3.7)	2.8 (2.3-3.7)	.98
Clopidogrel	3 (2.8)	0 (0)	Postreversal INR, median (IQR)	1.5 (1.2-1.8)	0.8 (0.7-1.0)	b.0001

Abbreviations: IQR, interquartile range; GCS, Glasgow Coma Scale; PTA, prior to admission.

^a Four patients were on warfarin for an unknown reason.

Time from reversal to INR 2.7 (1.5-4.7) 1.4 (0.5-2.3) b.0001 redraw (h), median (IQR)

hours of admission. In the postimplementation group, most patients re- ceived 3F-PCC (98.3%). A second dose of PCC was necessary in 8 patients (6.8%). The majority of both groups also received at least 1 dose of vita- min K (Table 3).

Prereversal INRs were similar among the groups (Table 3). Both re- versal to an INR of less than 1.4 with initial resuscitation and rebound INR were significantly improved in the postimplementation group (Table 3).

The incidence of expansion of hematoma on repeat imaging did not differ between preimplementation and postimplementation groups (P = .858) (Table 4). Transfusion requirements did not differ before or after reversal in either group (P = .306; P = .919) (Table 4).

Complications occurred at a significantly higher rate in the postimplementation group, with the exception of arterial thromboem- bolism and transfusion reactions, which were more common in the preimplementation group (Table 5). Unilateral Deep vein thrombosis was the most common complication in both groups, followed by superficial venous thrombosis, bilateral DVT, and pulseless electrical activity resulting in death attributed to extensive thrombosis (Table 5). In a subgroup analysis of thromboembolic events in patients with ICH, complications were again found to be more common in the post- implementation group (P = .006).

Baseline characteristic differences with a predefined P <= .2 were en- tered into a multivariable logistic fit for thrombotic complications. Base- line Glasgow Coma Scale and antiplatelet use were entered into and subsequently removed from the model, as those variables made the model unstable. Baseline SOFA score and having a mechanical valve as the indication for anticoagulation therapy were the only 2 factors

Table 2

Treatment characteristics

INR b1.4 after reversal, n (%) 43 (39.5) 104 (88.1) b.0001

Rebound INR N 1.4 in 48 h, n (%)^b 21 (48.8) 22 (21.2) .0013

^a Eight (6.8%) of postreversal patients required second dose of PCC.

^b Percentage of those who achieved INR less than 1.4.

independently associated with thrombotic complication in the nominal logistic regression analysis (P = .005) (Table 6).

Discussion

In this study, we showed that a reversal protocol consisting of weight-based 3F-PCC, fixed-dose rFVIIa, and vitamin K used to reverse warfarin-treated patients with life-threatening bleeds more consistent- ly reversed INR values to less than 1.4 as compared to the prior standard of care in a diverse population. This success came at the cost of a 2-fold increase in the risk of thromboembolic complications. The increased rate of adverse events was sustained even among serious complications, including death, ischemic stroke, or ST-elevation myocardial infarction, which occurred in 7 patients (6%) who received the reversal protocol, as compared to 1 patient (0.9%) in the standard of care group.

Current anticoagulation guidelines recommend the use of 4-factor PCC for urgent warfarin reversal [8,15]. Before the commercial availability of these products, institutions attempted warfarin reversal with 3F-PCC alone or combined with either FFP or rFVIIa to create a quasi-4-factor PCC. Edavettal et al [13] evaluated the use of 3F-PCC with vitamin K to correct INR in Geriatric patients presenting with traumatic ICH. Using a dose of 25 U/kg of Profilnine SD for patients with an INR greater than 1.5, their protocol resulted in significantly faster INR reversal as com- pared to prior standard of care reversal with FFP (P b .001). Three- factor PCC use was also associated with a decreased incidence of pro-

gression of ICH, without reported thromboembolic events [13].

Three-factor PCC combined with FFP and vitamin K has also been shown to rapidly normalize INR [11,14,16]. Siddiq et al [11] used a

Preimplementation (n = 109)

Postimplementation (n = 118)

Table 4

Clinical outcomes

Bleed location, n (%)

Head 87 (79.8)

Gastrointestinal 21 (19.3)

95 (80.5)

5 (4.2)

b.0001 1.0

.006

Preimplementation Postimplementation P

(n = 109) (n = 118)

Other^a 0 (0) 6 (5.1) Intervention/procedure, n (%) .10


Trauma	1 (0.9)	12 (10.2)	.003	Intracranial bleeds, n (%)	n = 87	n = 95

Surgery	27 (24.8)	27 (22.9)	Stable repeat imaging	61 (70.1)	69 (72.6)
interventional radiology	5 (4.6)	15 (12.7)	Transitioned to CMO before repeat	4 (4.6)	3 (3.2)

Expansion of bleed on repeat imaging

22 (25.3) 23 (24.2) .858

ICU LOS (d), median (IQR)	3 (2-5)	4 (2-6)	.07	Extracranial bleeds^a (U)	n = 22	n = 23
Hospital LOS (d), median (IQR)	5 (3-9)	7 (3-12)	.15	pRBC requirement before reversal	2 (0-4.25)	1 (0-7)	.306
Inhospital mortality, n (%)	16 (14.7)	25 (21.2)	.23	pRBC requirement after reversal	1 (0-2)	0 (0-2)	.919

Abbreviations: ICU, intensive care unit; LOS, length of stay.

^a Other bleed locations included thoracic and pericardial.

Abbreviation: CMO, comfort measures only.

^a Median (interquartile range).

Table 5

Complications

All patients, n (%)

Any complication^a	13 (11.9)	33 (27.9)	.003
DVT	3 (2.8)	16 (14.6)
Superficial thromboembolism	3 (2.8)	6 (5.1)
Bilateral DVT	3 (2.8)	5 (4.2)
PEA/death	1 (0.9)	4 (3.4)
NSTEMI	1 (0.9)	2 (1.7)
Ischemic stroke	0 (0)	2 (1.7)
arterial thrombosis	2 (1.8)	0 (0)
STEMI	0 (0)	1 (0.8)
Transfusion reaction	1 (0.9)	0 (0)
Intracranial bleeds, n (%) Any thromboembolic complication	9 (10.3)	25 (26.3)	.006

Preimplementation (n = 109)

Postimplementation P

(n = 118)

regression (P = .07). Although the thrombogenicity of rFVIIa is a known adverse effect and likely dose related, prior studies do suggest that some rFVIIa is necessary for effective anticoagulation reversal [12,16-19]. It is also interesting that 47.4% of the postimplementation group received FFP, with an average of 3.3 U, despite achieving the postreversal goal INR consistently.

Our study has several notable strengths, including large sample size, application to a diverse array of warfarin-related bleeding events, and external validity secondary to the elevated severity of illness and inclu- sion of elderly. There are also some notable limitations.

Limitations

Abbreviations: PEA, pulseless electrical activity; NSTEMI, non-ST-elevation myocardial in- farction; STEMI, ST-elevation myocardial infarction.

^a Patients may have experienced more than 1 complication.

Therapeutic dose of FFP (10-15 mL/kg) combined with a weight-based dose of Profilnine SD based on INR value (INR b 4, 25 U/kg; INR N 4, 50 U/kg). A higher percentage of patients treated with 3F-PCC, as com- pared to FFP with vitamin K, achieved target INR values within 4 hours (P = .012) [11].

Cabral et al [14] showed that similar dosing of 3F-PCC, but combined with fixed-dose FFP (2 U) and vitamin K, was also successful in correcting INR in warfarin-treated patients with ICH (P b .001). In the case series of 19 patients, 15.8% experienced a thromboembolic compli- cation [14]. Chapman et al [16] used a dose of 20 U/kg of 3F-PCC com- bined with FFP dosed at the physicians’ discretion, and when compared to FFP and vitamin K, 3F-PCC with FFP resulted in faster time to INR reversal (P = .048). Although the sample size was small, mortality was only seen in patients treated with 3F-PCC, and several sig- nificant thromboembolic events were noted [16].

When 3F-PCC was compared to both rFVIIa and FFP, all paired with vitamin K, it was found that INR normalization with FFP took twice as long compared to 3F-PCC or rFVIIa [10]. It was also noted that use of rFVIIa was also associated with more rebound INR (P = .001) [10]. No Thrombotic adverse events were noted in any group [10].

Our study revealed a surprisingly high number of thromboembolic complications. Although the 3F-PCC protocol reduced the time required for correction of INR significantly, it is possible that the protocol was too aggressive, as demonstrated by an average postreversal INR of 0.8 and increased thromboembolic complications.

Other published protocols using 3F-PCC dosed by INR may have pro- duced the lower thromboembolic complication rates due to utilization of a less aggressive INR-stratified reversal [11,14]. In our study, only 3 of the 34 patients who experienced a thromboembolic complication had an initial INR greater than 4. Therefore, had we used an INR- stratified dosing method, most patients who experienced a complica- tion would have been given a lower dose of PCC. It is also possible that the rFVIIa component of our protocol was unnecessary and potentially thrombogenic. Indeed, receipt of rFVIIa approached significance for as- sociation with thromboembolic complications in the multivariate

Table 6

Risk factor for thrombotic complications

Characteristic	P
SOFA score	.03
Receipt of rFVIIa	.07
Age	.28
AF as indication for anticoagulation	.47
Mechanical valve as indication for anticoagulation	.008
Traumatic injury as need for warfarin reversal	.11
GI hemorrhage as need for warfarin reversal	.49

Abbreviation: AF, atrial fibrillation.

Because of the retrospective nature of the study, it is prone to forms of bias.

The timing of the follow-up INR, although suggested in the reversal protocol, was left up to provider discretion, and therefore, time to INR reversal would be biased should a provider delay redraw of the INR postreversal.

It is likely that there were less severe bleeds in the preimplementation group, due to our inclusion criteria during this period of the study. These patients had lower SOFA scores and were identified via a different mechanism than the postimplementation group (use of ICD-9 coding). It is also possible that, in these patients, the provider’s goal could have been to substantially lower the INR value without necessarily correcting it to the reference range.

Although we show a surprisingly high thromboembolic complica- tion rate with the reversal protocol, it is not possible to directly link the event to PCC administration, given the severity of illness in our pa- tients and varying lengths of time between receipt of the reversal proto- col and the complication. A significant proportion of patients in the postimplementation group received FFP in addition to the reversal pro- tocol, which could have further attributed to thromboembolic events.

Conclusion

A reversal protocol using weight-based 3F-PCC, fixed-dose rFVIIa, and vitamin K appears to be effective in normalizing INR values in a di- verse array of bleeding events. Caution should be used, and potential dose reductions, considered, when using this protocol due to elevated risk of thrombotic complications.

References

AJEM