A proposal for managing bleeding in patients on therapeutic factor XI(a) inhibitors

Abstract

Several drugs that reduce functional levels of the plasma protease zymogen factor XI (FXI), or that inhibit its activated form (FXIa), are being evaluated as treatments to prevent thrombosis. Based on the observation that individuals with inherited FXI deficiency have a relatively mild bleeding disorder, it is anticipated that therapeutic FXI(a) inhibitors will have a smaller impact on hemostasis than anticoagulants targeting thrombin or factor Xa. However, even if FXI(a) inhibitors are determined to be safer than currently used anticoagulants, some patients on these drugs will experience abnormal bleeding or require emergent surgery. Strategies for dealing with such situations are required. Treatment with antifibrinolytic agents and low doses of recombinant factor VIIa effectively prevent abnormal bleeding in FXI-deficient patients with alloantibody inhibitors to FXI who undergo surgery. We propose that a similar strategy can be used for patients on therapeutic FXI(a) inhibitors who are bleeding or require invasive procedures.

1 |. INTRODUCTION

Anticoagulants currently used to treat or prevent thromboembolism directly or indirectly target the active sites of the plasma coagulation proteases thrombin and/or factor Xa (e.g., heparins, fondaparinux, bivalirudin, argatroban, direct oral anticoagulants [DOACs]), or lower the plasma concentrations of the precursors of these enzymes (e.g., vitamin K antagonists). Because thrombin and factor Xa are central to the hemostatic response to injury, these drugs increase a patient’s risk for severe bleeding, including life-threatening intracranial and gastrointestinal hemorrhage.^1–4 This reality has led to development of strategies for preventing or stopping anticoagulant-associated bleeding based on neutralizing or bypassing the anticoagulant^2,5,6 or, in the case of vitamin K antagonists, reconstituting normal levels of coagulation factors.^7,8

There has been substantial interest over the past 2 decades in identifying targets for antithrombotic drugs that play a smaller role in hemostasis than do thrombin and factor Xa. Inhibition of such targets is expected to produce safer anticoagulation from the standpoint of therapy-induced bleeding. Factor XI (FXI), the precursor of the serine protease factor XIa (FXIa), has shown promise in this regard.^9–11 Inhibition or reduction of FXIa produces a potent antithrombotic effect in preclinical models,^12–15 and data from clinical trials reveal an impressive capacity to reduce venous thrombus formation in patients undergoing total knee arthroplasty (TKA).^16–18 The mild bleeding tendency observed in FXI-deficient patients,^19,20 and in patients on FXI inhibitors in the TKA trials,^16–18 suggest that agents directed against FXI or FXIa may produce a useful antithrombotic effect without significantly compromising hemostasis. But even if FXIa inhibition is a safer method for anticoagulation, bleeding will inevitably occur, and emergent invasive procedures will be required, in some patients taking these drugs. As with DOACs targeting thrombin and factor Xa, strategies will be needed to deal with these situations. We propose that lessons learned from managing FXI-deficient patients with FXI alloantibody inhibitors will be applicable to managing patients on therapeutic FXIa inhibitors.

2 |. BLEEDING IN FACTOR XI-DEFICIENT PATIENTS

In the activated partial thromboplastin time (APTT) assay, FXIa drives clot formation by activating factor IX. Accordingly, plasmas of patients with inherited FXI deficiency clot slowly in APTT assays. Indeed, the APTT is usually prolonged to a greater extent by complete deficiency of FXI than by deficiencies of factors VIIII or factor IX, the proteins missing in patients with the severe bleeding disorders hemophilia A and hemophilia B, respectively. Yet FXI deficiency does not cause a hemophilia-like syndrome, and the propensity of FXI-deficient patients to bleed correlates poorly with the APTT.^19–22 This may be explained by observations indicating that the major activator of factor IX during hemostasis is likely to be the factor VIIa/tissue factor complex, and not FXIa.^23,24 It is clear that the critical role that FXIa plays in clotting in the APTT assay does not reflect its more limited contribution to hemostasis at an injury site.

The bleeding disorder associated with FXI deficiency is considerably milder than in hemophilia.^19–22 Indeed, most FXI-deficient patients do not experience abnormal hemostasis. The hemarthroses and soft-tissue bleeds that are hallmarks of hemophilia, as well as intracranial hemorrhage and gastrointestinal bleeding, are not features of severe FXI deficiency. Abnormal hemostasis in FXI-deficient individuals typically occurs with trauma or surgery, and “spontaneous” bleeding (with the exception of heavy menstrual bleeding in some woman) is unusual. There is a predilection for excessive bleeding when injury involves the nasopharynx, mouth, and urinary tract that is attributed to high fibrinolytic activity in these tissues. Bleeding with trauma to other areas is less consistent, and procedures such as knee replacement, appendectomy, simple cataract removal, cesarean section, circumcision, and tooth extraction are usually well-tolerated without factor replacement.^25–28

3 |. TREATING FACTOR XI-DEFICIENT PATIENTS WITH ANTI-FACTOR XI ALLOANTIBODY INHIBITORS

Patients with FXI deficiency undergoing major surgery have traditionally received factor replacement, alone or in combination with fibrinolytic inhibitors such as tranexamic acid (Cyklokapron, Lysteda) or ε-aminocaproic acid (Amicar).^19,20 Replacement-based strategies using fresh frozen plasma (FFP) or FXI concentrate are effective but have some drawbacks. FFP is the replacement option available for FXI-deficient patients in most of the world. Limits on the volume of FFP that can be infused over a short period place restrictions on the amount of factor that can be administered in many situations. FXI concentrates are available in some European countries and in Canada.²⁹ Use of the early versions of these concentrates could be complicated by thrombotic events (mostly arterial) that usually occurred after administering large doses to patients with other risk factors for thrombosis.^30,31 Traces of FXIa in the concentrates likely contributed to thrombotic events, and current preparations contain heparin and serpins such as antithrombin and C1-inhibitor to neutralize FXIa. Although these steps appear to make FXI concentrates safer, the preparations are still considered by some to be prothrombotic.³²

A second issue of concern with factor replacement is development of alloantibody inhibitors to FXI. A number of mutations that cause complete deficiency of plasma FXI antigen in humans have been described, including FXI Glu117Stop, which has an allele frequency of >2% in individuals of Ashkenazi Jewish ancestry.³³ Homozygotes for Glu117Stop account for ~25% of patients with severe FXI deficient in Israel. Previously, we reported that about one-third of these patients develop neutralizing alloantibody inhibitors to FXI after factor replacement, often after a single exposure to FXI.³⁴ The inhibitors usually do not cause bleeding per se, but do interfere with the effectiveness of factor replacement when it is required. This can lead to treatment failure during surgery, and is particularly problematic when the presence of the inhibitor is not suspected.

An effective strategy developed for managing patients with FXI alloantibodies who require surgery is to combine antifibrinolytic therapy and a low dose of a “bypassing agent” such as recombinant factor VIIa (rFVIIa).^35–37 This approach has been used successfully in a variety of surgical settings, including repair of a dissecting aneurysm of the ascending aorta, and aortic valve replacement.^37–40 The strategy was employed in the following case:

3.1 |. Case 1

A 76-year-old man with severe FXI deficiency (<1% of normal plasma activity, homozygous for F11 Glu117Stop) was diagnosed with adenocarcinoma of the pancreas. There was no evidence of metastatic disease on positron emission tomography-computed tomography scan. The patient had received FFP as prophylaxis before prior surgical procedures and wisdom tooth extraction and subsequently developed an alloantibody FXI-inhibitor. The titer of the inhibitor was 5 Bethesda Units at the time his cancer was diagnosed. A Whipple procedure (en bloc removal of the distal stomach, duodenum, head of the pancreas, common bile duct, and gallbladder) was planned. Because the tumor extended into the distal pancreas, the procedure was expanded to include a total pancreatectomy and splenectomy. The patient received 1500 mg tranexamic acid (body mass 110 kg) immediately before incision. At the end of surgery, a single bolus of rFVIIa (10 μg/kg) was administered, and tranexamic acid was continued at 1500 mg every 6 h. Twenty-four hours after surgery, the treatment was changed to tranexamic acid 1000 mg every 6 h for 5 additional days. There was no evidence of excessive bleeding on this regimen.

4 |. NONREPLACEMENT THERAPY IN FXI-DEFICIENT PATIENTS WITHOUT INHIBITORS

Several issues surrounding FXI replacement, including the thrombogenic potential and alloantibody inhibitor development, and the experience with alternative therapies in patients with alloantibodies, have led to changes in the manner in which FXI-deficient patients without alloantibodies are managed at some institutions. Specifically, strategies based on antifibrinolytic drugs with or without small doses of rFVIIa have become the preferred frontline therapy for covering FXI-deficient patients during surgery,^38,39,41 as illustrated by the following case.

4.1 |. Case 2

A 42-year-old woman was scheduled for partial nephrectomy to remove a 3.6 × 4 cm mass in the lower pole of the right kidney that was suspected to be malignant. Positron emission tomography- computed tomography scans did not show evidence of metastasis. The patient did not have a history of abnormal bleeding, but during preoperative evaluation her APTT was noted to be prolonged (54 s, normal 24–33 s). Further evaluation identified low FXI activity (4% of normal) and genetic analysis revealed compound heterozygosity for two F11 mutations, Glu117Stop and Leu283Phe. This is the most common genotype associated with severe FXI deficiency in Israel. A robotic partial nephrectomy was performed. The patient received tranexamic acid 1000 mg before skin incision and then every 6 h for 1 week. At the end of surgery, she received a single dose of rFVIIa (10 μg/kg). There was no evidence of excessive bleeding during or after surgery. The mass was subsequently confirmed to be renal cell carcinoma.

Combining antifibrinolytics with rFVIIa has the potential to increase thrombotic risk. However, adequate hemostasis may be achieved in FXI-deficient patients with amounts of rFVIIa that are well below the repeated high doses (≥90 μg/kg) used to treat patients with factor VIII inhibitors.^38,39,41 Thrombosis is rare with doses of rFVIIa <30 μg/kg.⁴² We recently reported that tranexamic acid, 1 g before surgery and then 1 g every 6 h for 3–5 days after surgery, supplemented by a single dose of rFVIIa (10–15 μg/kg) given at the end of surgery, prevented abnormal bleeding in patients undergoing a variety of surgical procedures.⁴¹

Our experience and recommendations for managing FXI-deficient patients through surgery are summarized in Table 1. It is our impression that antifibrinolytic therapy is the key to reducing bleeding in FXI-deficient patients in the absence of factor replacement. This is in keeping with the observation that these patients bleed most often, and most severely, when trauma involves tissues with high intrinsic levels of fibrinolytic activity.^19,20 In these tissues, thrombin generation through FXIa, in addition to promoting fibrin formation and platelet activation, may increase activation of the metalloprotease thrombin-activatable fibrinolysis inhibitor, which modifies fibrin to increase its resistance to fibrinolytic degradation.⁴³

TABLE 1.

Suggested prophylactic regiments for patients on FXI(a) inhibitors requiring surgery

Low risk	Moderate risk	High risk
Observation or antifibrinolytic treatmenta,b	Antifibrinolytic treatmenta,b	Antifibrinolytic treatment and recombinant factor VIIaa,b
Cardiac catheterization	Oral and ENT surgery	Cardiac surgery
Cataract surgery	Periodontal surgery	Aortic aneurysm repair
PICC line placement	Transurethral prostatectomy	Pulmonary surgery
Fracture fixation	Lithotripsy	Mediastinoscopy
Carpal tunnel surgery	Tonsillectomy	Brain/CNS surgery
Endoscopy without biopsy	Rhinoplasty	Partial or total nephrectomy
Colonoscopy without biopsy	Tooth extraction	Colectomy
Skin biopsy	Lymph node biopsy	Total gastrectomy
Appendectomy	Pacemaker insertion	Pancreatectomy
	Defibrillator insertion	Thyroidectomy
	Vagal nerve stimulation	Cystectomy
	Deep skin resection	Facelift
	Mastectomy	Radical prostatectomy
	Inguinal hernia repair	Laparoscopic cholecystectomy
	Ventral hernia repair	Major neck surgery
	Sleeve gastrectomy
	Liposuction
	Vitrectomy
	Colonoscopy with biopsy
	Hysterectomy with biopsy

Abbreviations: CNS, central nervous system; ENT, ear-nose-throat; PICC, peripherally inserted central catheter.

^aIn situations in which surgery involves tissues with high fibrinolytic activity, administration of antifibrinolytic drugs may be extended to 10 days.

^bDoses of antifibrinolytic drugs may need to be adjusted in patients with reduced renal function.

The role of rFVIIa in our approach is less clear. The efficacy of such small amounts of rFVIIa could reflect the modest contribution of FXIa to thrombin generation during hemostasis in most tissues. In the absence of FXI, rFVIIa may promote hemostasis by enhancing fibrin resistance to fibrinolysis through formation of more tightly packed fibrin fibers,⁴⁴ as well as through thrombin-activatable fibrinolysis inhibitor activation. Furthermore, rFVIIa may cleave protease-activated receptor 1 to induce extracellular vesicle release that further potentiates thrombin generation.⁴⁵ However, we suspect that low-dose rFVIIa may appear to be so effective partly because antifibrinolytic therapy alone is adequate for achieving hemostasis in many cases. This is consistent with the impression that FXI primarily counters fibrinolysis, and that its importance is diminished when fibrinolysis is inhibited. Low-dose rFVIIa, then, may function primarily as a backstop in our strategy that deals with bleeding that breaks through antifibrinolytic therapy. Hypothetically, the need for rFVIIa might be circumvented by increasing the dose of the antifibrinolytic agent. We have used this strategy in some cases. However, the approach should be used with caution because of the risk of drug-induced kidney injury and seizures at high doses.⁴⁶

5 |. THER APEUTIC FACTOR XI(a) INHIBITORS FOR PREVENTING THROMBOSIS

DOACs directed against thrombin and factor Xa were approved for use without the availability of specific agents to neutralize their effects. Because of their impact on hemostasis, these drugs need to have relatively short half-lives (on the order of hours) as safety features. This issue may be of less concern for drugs directed at FXIa if, as predicted, they do not cause severe spontaneous bleeding. Several compounds that reduce FXI activity for days to weeks have been tested in phase 2 trials (Table 2). They include an antisense oligonucleotide (IONIS-FXI_Rx or BAY2306001) that decreases plasma FXI protein by specifically reducing FXI synthesis in the liver,¹⁶ antibodies (AB023⁴⁷ and abelacimab¹⁸) that bind zymogen FXI and interfere with its activation and activity, and an antibody (osocimab) that inhibits the FXIa active site.¹⁷ IONIS-FXI_Rx, osocimab, and abelacimab were comparable or superior to standard-dose enoxaparin for preventing deep vein thrombosis in patients undergoing TKA. In this setting, the drugs did not compromise hemostasis significantly. Indeed, IONIS-FXI_Rx and osocimab administered preoperatively did not produce an obvious detrimental effect on hemostasis during or after TKA.^16,17 AB023 was given to patients with end-stage kidney disease on chronic hemodialysis⁴⁷ and reduced clot formation on dialyzer membranes without increasing bleeding.

TABLE 2.

Drugs targeting FXI(a) undergoing testing in clinical trials

Drug	Drug type	Mechanism of action	Administration route	Effective half-life
IONIS-FXIRx	DNA antisense oligonucleotide	Reduces level of FXI antigen in plasma	Subcutaneous	Days to weeks
Osocimab	Monoclonal antibody	FXIa active site inhibitor	Intravenous or subcutaneous	Days
AB023	Monoclonal antibody	Binds FXI and interferes with conversion to FXIa	Intravenous or subcutaneous	Days
Abelacimab	Monoclonal antibody	Binds FXI, interferes with conversion to FXIa and inhibits FXIa activity	Intravenous or subcutaneous	Days
Milvexian	Small molecule	FXIa active site inhibitor	Oral	Hours
BAY-243334	Small molecule	FXIa active site inhibitor	Oral	Hours

Abbreviations: FXI, factor XI; FXIa, activated form of factor XI.

6 |. PREVENTING BLEEDING IN PATIENTS ON THER APEUTIC FXI(a) INHIBITORS

Although our limited experience with drugs targeting FXI suggest they are safe when administered over short periods (effect lasting days to a few weeks), it is conceivable that bleeding will be a greater problem with longer drug exposures, or when the drugs are given to patients with comorbidities that adversely affect hemostasis. Furthermore, patients on these agents may require emergent surgery or suffer life-threatening bleeding for reasons that may be unrelated to the FXI inhibitor. In formulating approaches to deal with such situations, it seems reasonable to start by considering currently available treatments, and their likely effects on FXI(a) inhibitors with different mechanisms of action.

FXI replacement with FFP or FXI concentrate should reverse the effect of IONIS-FXI_Rx, which causes a true plasma deficiency of FXI antigen. Factor replacement could have a role in neutralizing anti-FXI antibodies that bind zymogen FXI, such as AB023 and abelacimab. Hypothetically, it should be possible to saturate these antibodies with the target protein. Any FXI beyond what is required for saturation would be available to contribute to hemostasis. However, the practicality of such a strategy would be dictated by the concentration of unbound antibody requiring neutralization. In many cases, the amount of FXI needed to achieve saturation may be impractically large. Furthermore, overshooting the required dose could lead to high plasma concentrations of free FXI, increasing thrombosis risk. The effects of inhibitors directed against the FXIa active site, including the antibody osocimab and the small molecule oral FXIa inhibitors milvexian (JNJ70033093/BMS-986177)^48,49 and BAY-243334,⁵⁰ may not be altered appreciably by FXI replacement, because functional FXI (as opposed to FXIa) is not low in patients on these drugs. One could design specific neutralizing agents for FXIa active site inhibitors, as has been done for DOACs targeting factor Xa and thrombin.^5,6 Indeed, properties of an antibody-based reversal agent for milvexian were recently reported.⁵¹ However, we feel that specific reversal agents are not necessary for drugs targeting FXI(a).

In our opinion, a practical approach for preparing patients on FXI(a) inhibitors for surgery would be to adopt the strategy described here based on the use of antifibrinolytics and rFVIIa in FXI-deficient patients with or without alloantibodies. Before surgery, we recommend discontinuing antiplatelet agents, if possible, and addressing any coagulopathies distinct from FXI deficiency. Antifibrinolytics and rFVIIa could be used for patients on any of the FXI(a)-targeting drugs currently being tested. In support of this approach, it was recently demonstrated that low-dose rFVIIa effectively reverses the effects of milvexian in a rabbit AV-shunt model of thrombosis.⁵²

7 |. TREATING BLEEDING IN PATIENTS ON FXI(a) INHIBITORS

When considering options for dealing with acute bleeding in patients with FXI deficiency, it is important to keep in mind that bleeding in this disorder is usually not severe, and that most FXI-deficient individuals do not exhibit abnormal hemostasis. The key implication of this is that bleeding in a FXI-deficient patient may not be caused by, or even be influenced by, the factor deficiency. Agents such as antifibrinolytic drugs and rFVIIa that enhance hemostasis through mechanisms that do not involve FXI may, therefore, be more effective than FXI replacement for arresting bleeding in many situations. It seems reasonable to conclude that the same consideration will apply to patients on therapeutic FXI(a) inhibitors, and that antifibrinolytic drugs and rFVIIa may be more effective at stopping bleeding than specific agents that neutralize a FXI(a) inhibitor.

Non-life-threatening bleeding in FXI-deficient patients usually responds to local measures and antifibrinolytic therapy. For severe or life-threatening bleeding in patients on a FXI(a) inhibitor, including intracranial or gastrointestinal hemorrhage, rupture of the spleen, or a dissecting aneurysm of a major vessel, we recommend antifibrinolytic agents combined with rFVIIa. The APTT is expected to be prolonged in patients taking FXI inhibitors, but a prolonged prothrombin time could indicate a coagulopathy unrelated to the inhibitor. Administration of FFP or prothrombin complex concentrate (PCC) should be considered in these cases. Plasma exchange to remove an inhibitor directed at FXIa would be an option in a patient with bleeding that is refractory to other treatment, although the efficacy of (or need for) such an approach is unknown.

We appreciate that rFVIIa is not available in most hospitals, and its availability even at centers that specialize in care of patients with hemophilia may decrease as new treatments such as emicizumab become widely adopted. For elective procedures, transfer to a facility that uses rFVIIa is an option, but other strategies may be needed for timely intervention for acute bleeding. An activated PCC such as FEIBA could be used in place of rFVIIa, although we are unaware that this has been tried in FXI-deficient patients. FEIBA’s thrombogenic potential must be considered, particularly when combined with antifibrinolytic drugs, and initial use of small doses may be prudent. Nonactivated three- and four-factor PCCs have been used to treat emergent bleeding in a variety of situations and could be tried in bleeding patients on a factor XI(a) inhibitor. But, again, there is little track record for their use specifically in FXI-deficient patients.

8 |. CONCLUSIONS

Over the past 60 years, hands-on experience with patients lacking FXI has provided important insight into the biology of this enigmatic clotting factor. Factor XI-deficient patients experience milder and more unpredictable bleeding than do patients with severe hemophilia A or hemophilia B, despite the comparable effects on commonly used coagulation tests such as the aPTT. There is mounting evidence that FXI-deficient patients can be managed during surgery with strategies that do not involve factor replacement. In our opinion, it is now rarely necessary to replace the missing protein in FXI-deficient patients. We propose that strategies based on antifibrinolytic drugs and rFVIIa that are currently employed in treating patients with inherited FXI deficiency (with or without inhibitors) will be effective for patients on therapeutic FXI inhibitors with bleeding or who require emergent surgery.