Efficacy of pro-haemostatic agents in the management of factor Xa inhibitor-associated intracranial haemorrhages

Introduction

Direct oral anticoagulants (DOACs) are now the recommended anticoagulants for the treatment and secondary prevention of venous thromboembolism (VTE), thromboprophylaxis after major orthopaedic procedures and the prevention of stroke and systemic embolization in non-valvular atrial fibrillation. DOACs that are approved for these indications in Australia include direct thrombin inhibitors (dabigatran (Pradaxa)) and direct factor Xa inhibitors (apixaban (Eliquis) and rivaroxaban (Xarelto)). They have been shown to be as efficacious as vitamin K antagonists (VKAs) with many superior features, such as a better safety profile, fewer drug and food interactions, rapid onset of action and ease of administration. However, unlike VKAs, options for reversing DOACs are limited and are of particular concern in situations such as severe bleeding.

The incidence of intracranial haemorrhage (ICH) in patients taking a DOAC is 0.1%–0.2%, with a 30- to 90-day mortality rate of 40% to 65%.^{1, 2} A meta-analysis comparing the risk of ICH between DOACs and VKAs has shown that DOACs are associated with a lower risk of ICH (relative risk: 0.40–0.59).³ Patients receiving DOACs also have less severe ICHs and lower mortality rates, likely due to lower volume haemorrhages at admission and less frequent haematoma expansion.⁴ However, given the high morbidity and mortality associated with ICHs, the management of patients with anticoagulation-associated ICHs needs to be optimised with the reversal of anticoagulant activity at the forefront of management.

Although there are several guidelines for the management of DOAC-associated bleeding, evidence from randomised controlled studies is lacking for specific reversal strategies. Pro-haemostatic agents such as prothrombin complex concentrates (PCCs) have been used in the event of life-threatening bleeding associated with rivaroxaban and apixaban. Their use poses the risk of thrombotic complications, especially in a population of patients who have already proven themselves to be procoagulant. In animal models of ICH, PCC was more effective than fresh frozen plasma (FFP) at preventing haematoma expansion.⁵ There is still limited evidence in their efficacy in patients who present with DOAC-associated life-threatening bleeding.⁶

We performed a retrospective audit of rivaroxaban and apixaban-associated intracranial haemorrhages that presented to our local health district between 2015 and 2021 to determine the efficacy of pro-haemostatic agents in these patients.

Methods

This study was approved by the South West Sydney Local Health District (SWSLHD) Human Research Ethics Committee. Through our laboratory database (PathNet), we identified patients who had apixaban and rivaroxaban levels performed at Liverpool Hospital Coagulation Laboratory from January 2015 to December 2021 inclusive. This included inpatients and outpatients attending Liverpool Hospital (a tertiary trauma hospital in metropolitan Sydney) and its five referring hospitals: Bankstown/Lidcombe Hospital, Fairfield Hospital, Campbelltown Hospital and Bowral Hospital; constituting over 1.1 million residents, approximately 13.4% of the New South Wales population.

All assays were performed on the Stago STA Evolution haemostasis analyser and performed in accordance with manufacturer instructions. All drug assays were performed on site at Liverpool Hospital Coagulation Laboratory. DOAC levels resulted at a median time of 45 minutes (range 17–49 minutes) if the patient was managed at Liverpool Hospital. The median specimen transport time for referring hospitals was 70 minutes (range 66–200 minutes). Apixaban and rivaroxaban levels were measured using their respective STA-Liquid Anti-Xa calibrator and control kits. A DOAC level of <30 ng/mL is considered clinically insignificant in our institution, i.e. not contributing to excess bleeding.

Data were then extracted retrospectively through the electronic record system by two authors (SN, DN). The following information was collected: patient age, gender, weight, creatinine clearance, DOAC dosing schedule, indication(s) for anticoagulation, indication(s) for measurement, DOAC level, concomitant antiplatelet therapy, volume of intracranial haemorrhage (see Sectra volume measurement tool below), grade of bleeding, HAS-BLED score, CHADs-VASc score (if history of atrial fibrillation), clinical management of these patients including the use of reversal agents (if indicated), in-hospital outcomes (including discharge destinations) and ongoing anticoagulation decisions.

The ICH volume was calculated using the Sectra volume measurement tool, which is an automatic volume measurement. This is calculated by assessing the difference in computed tomography (CT) number (Hounsfield units) between the haematoma and the surrounding brain parenchyma. The CT number corresponds to the radiodensity of the tissue, which is higher for a haematoma compared to the surrounding brain parenchyma.⁷ To measure the volume using this tool, a multiplanar reconstruction (MPR) is created using the volumetric data of the scan to create a three-dimensional image by converting the data acquired in a certain imaging plane into another plane. Once the MPR is performed, the volume tool is activated to measure a volume by delineating the region of interest. Though the Sectra is an automated tool, radiologists can adjust the dimensions measured in all three planes to improve accuracy, primarily in situations where the object of interest is irregular and not uniform. This software uses the generic mathematical equation for ellipsoid objects (V = length × width × depth × 0.532) and provides the volume in mm³. This method has been proven to be as accurate as manual volume measurement and is faster and more efficient.⁸

Results

During this study period, a total 1644 apixaban levels and 1317 rivaroxaban levels were performed with a total of 688 patient encounters presenting with major or clinically relevant non-major bleeding. Of these, 342 patients (49.7%) were classified as having major and/or life-threatening bleeding (World Health Organization grade 3/4).⁹ A total of 150 patients presented with ICH, accounting for 43.9% of major bleeding. Ninety-four of these were spontaneous ICHs (63%) and the rest occurred in the context of trauma. Approximately 66% of these ICHs were in patients receiving apixaban (99/150) and 34% were in those receiving rivaroxaban (51/150).

Patient demographics and clinical characteristics are shown in Table 1. The median age was 81 years (interquartile range (IQR): 72–86 years) in the apixaban group and 76 years (IQR: 71–79 years) in the rivaroxaban group, with 52% male patients in the apixaban group and 65% in the rivaroxaban group. The primary indication for anticoagulation was for atrial fibrillation (81% apixaban, 67% rivaroxaban). Approximately 20% of patients were taking concomitant antiplatelet therapy, with three patients taking dual antiplatelet therapy (aspirin and clopidogrel). Patients with apixaban-related bleed presented with an average creatinine clearance of 57.5 mL/min, whereas the average creatinine clearance for patients taking rivaroxaban was 71.9 mL/min. A total of eight patients taking apixaban and one patient taking rivaroxaban presented with acute renal insufficiency (<25 mL/min for apixaban and <15 mL/min for rivaroxaban). The median CHADsVASc score was 4 for patients taking apixaban and 3 for patients taking rivaroxaban, and the HAS-BLED score was 2 in both groups.

The median apixaban level was 93 ng/mL and the median rivaroxaban level was 62 ng/mL on presentation. Thirty-five of the 150 (23%) patients who presented with ICH had DOAC levels <30 ng/mL. Six patients underwent surgical drainage of their ICH (4%).

A total of 96 patients (64%) received pro-haemostatic agents (either FFP, Prothrombinex-VF, factor eight bypassing agent (FEIBA) or a combination of all three agents; as per the 2013 Clinical Excellence Commission (CEC) guidelines): 70% in the apixaban group, 53% in the rivaroxaban group. Fifteen of these patients had DOAC levels <30 ng/mL (15/96, 16%). There were no in-hospital thrombotic complications in the patients who received pro-haemostatic agents.

Discussion

Patients presenting with factor Xa (FXa) inhibitor-associated bleeding were older (≥75 years), had mild renal impairment on presentation and were predominantly on anticoagulation for a history of atrial fibrillation (76%), with an average CHADsVASc score of 4 and HAS-BLED score of 2.

The 2023 CEC DOAC guidelines recommend that the standard of care for reversal of FXa inhibitors is with andexanet alfa (Andexxa), and comments that PCC can be considered if andexanet alfa is not available.¹⁰ Andexanet alfa was approved by the Australian Therapeutic Goods Administration (TGA) in July 2023 for the reversal of FX inhibitors in patients who present with major or life-threatening bleeding. Andexanet alfa is a recombinant modified human FXa decoy protein. It acts by binding to FXa inhibitors with affinities similar to that of native FXa, serving as a decoy that sequesters FXa inhibitors until they can be cleared. ANNEXA-A and ANNEXA-R demonstrated the reversal of the anticoagulant activity (apixaban and rivaroxaban respectively) within minutes after administration and for the duration of infusion without evidence of clinical toxic effects.¹¹

During this study period, only Prothrombinex-VF was available in Australia. Prothrombinex-VF contains purified human coagulation factors II, IX and X and low levels of factors V and VII. Precautions to use include a risk of thrombosis, embolism, disseminated intravascular coagulation or myocardial infarction, which may be fatal. Prothrombin complex concentrate is indicated in patients with deficiency caused by treatment with VKAs (warfarin) or in patients with single or multiple congenital deficiencies of factor IX, II or X. Its use in patients with FXa inhibitor-related bleeding is off-label with limited evidence.

The use of pro-haemostatic agents in our patient cohort did not significantly alter the mortality rate (29% vs 22%) of bleeding patients in our study. This is similar to population studies with in-hospital and 1-year mortality rates of 32.4%¹² and 46%¹³ respectively.

The use of Prothrombinex-VF has been reported to be associated with a composite arterial/venous thromboembolic rate of 4.6% within 30 days.¹⁴ Our study did not identify an increased rate of thrombosis associated with its use despite the high percentage of AF patients. There were no documented in-hospital thrombotic events in our cohort of patients who received pro-haemostatic agents.

Patients who presented with clinically significant DOAC levels (>30 ng/mL) had a higher risk of mortality that did not appear to change with the use of pro-haemostatic agents.

Patients with a GCS score <6 had high mortality rates reported at 42–87.5%.¹⁵ In our study, we had 14 patients who presented with a GCS score <7. All patients died except for one patient who had a GCS score of 3 requiring intubation likely due to alcohol intoxication rather than to his ICH.

Guidelines for the use of andexanet alfa in ICH patients have been derived from ANNEXA-I (exclusion criteria: GCS score <7, National Institutes of Health Stroke Scale score >35, ICH volume >60 mL, life expectance <1 month, thrombotic events within 2 weeks) to avoid its inappropriate use.¹⁶ Reliable and prompt measurement of on-board anticoagulant activity would further reduce the risk of patients inappropriately receiving andexanet alfa and the associated significant financial burden. A delay in readily available DOAC levels was associated with the inappropriate use of pro-haemostatic agents in 10.3% of patients in our study, which not only places patients at potential harm but also translates to an inappropriate use of resources and associated financial burden.

Current anti-FXa activity assays are unsuitable for measurement following administration of andexanet alfa. This is because of high sample dilution, which results in andexanet alfa/FXa inhibitor dissociation that inaccurately gives high FXa inhibitor-measured concentrations. Point-of-care testing can potentially be used to measure residual apixaban or rivaroxaban activity following administration of andexanet alfa.¹⁷

A strength of our study is the granularity of data with regards to patient management and outcomes that followed with correlation to DOAC levels. Limitations of our study include the small sample size and that not all patients who present to our LHD with major and/or life-threatening bleeding have been captured, as only those who have had DOAC assays performed are captured through our audit. Although it is standard of care for DOAC assays to be ordered when a patient presents to our LHD with DOAC-associated bleeding, a later internal audit of the total ICH cases presenting to our LHD showed only 21% of these patients had a DOAC level ordered (unpublished). There is also a potential for treatment bias by the treating clinicians for those who did or did not receive reversal agents. Another limitation of our study is that the imaging series is incomplete, primarily in patients who were deemed to have a further management that was futile. The expansion of intracranial haemorrhages is highest in the first few hours of onset of bleeding, and the benefit of reversal of anticoagulant activity, alongside supportive measures such as blood pressure control, is likely during this window. The time to presentation and management could not be reliably determined for each patient, which is another limitation.

Conclusion

Patients presenting with DOAC-associated ICH have a higher risk of mortality that does not appear to change with the use of pro-haemostatic agents, namely FFP, Prothrombinex-VF and FEIBA in our cohort of patients. The identification of appropriate patients who might benefit most from pro-haemostatic agents and DOAC reversal agents such as andexanet alfa is essential provided that this can be performed in a timely and accurate manner to reduce overall morbidity and mortality.