Fibrin Clot Permeability as a Predictor of Stroke and Bleeding in Anticoagulated Patients With Atrial Fibrillation
Background and Purpose—Formation of denser fiber networks has been reported in atrial fibrillation and ischemic stroke. In this longitudinal cohort study, we evaluated whether fibrin clot density may predict thromboembolic and bleeding risk in patients with atrial fibrillation on vitamin K antagonists.
Methods—In 236 patients with atrial fibrillation receiving vitamin K antagonists treatment, we measured ex vivo plasma clot permeability (Ks), a measure of the pore size in fibrin networks.
Results—During a median follow-up of 4.3 (interquartile range, 3.7–4.8) years, annual rates of ischemic stroke or transient ischemic attack and major bleeds were 2.96% and 3.45%, respectively. In multivariate Cox regression analysis, patients with lower Ks (<6.8 cm2×10−9, median) had increased risk of ischemic stroke or transient ischemic attack (hazard ratio [HR], 6.55; 95% confidence interval [CI], 2.17–19.82) and major bleeds (HR, 10.65; 95% CI, 3.52–32.22). Patients with elevated Ks (≥6.8 cm2×10−9) had an increased rate of minor bleeding compared with the remainder (11.63% per year versus 3.55% per year; P<0.0001). The independent predictors of stroke or transient ischemic attack were low Ks (<6.8 cm2×10−9; HR, 7.24; 95% CI, 2.53–20.76), age (HR, 1.05; 95% CI, 1.01–1.10), and treatment with angiotensin-converting enzyme inhibitors (HR, 2.27; 95% CI, 1.08–4.77). Major bleeds were predicted by low Ks (<6.8 cm2×10−9; HR, 8.48; 95% CI, 2.99–24.1) and HAS-BLED score ≥3 (HR, 2.22; 95% CI, 1.12–4.38).
Conclusions—This study is the first to show that unfavorable fibrin properties reflected by formation of denser fibrin networks determine, in part, the efficacy and safety of anticoagulation with vitamin K antagonists in patients with atrial fibrillation.
Oral anticoagulant therapy is effective in the prevention of thromboembolic events by ≈64% in patients with atrial fibrillation (AF) at the cost of major bleeding occurring at a rate of 2% to 4% per year.1 Most risk stratification models for predicting stroke, systemic embolism, and bleeding events in patients with AF, including CHA2DS2-VASc score, are based solely on clinical risk factors. Several blood biomarkers, reflecting cardiac and renal function, oxidative stress, and inflammation, have been reported as independent predictors of clinical outcome during anticoagulation in AF, for example, cardiac troponin or cystatin C.2 Interestingly, the overlap between ischemic stroke and bleeding risk factors is common.3
Fibrin is a major component of both venous and arterial thrombi, including those retrieved in stroke and myocardial infarction by mechanical thrombectomy.4 Fibrin clot structure affects the transport of proteins involved in fibrinolysis through networks and largely determines the fibrinolysis rate with fast fibrinolysis of fibrin clots composed of loosely packed fibers.5,6 Previous studies have shown that hypercoagulable state occurs in patients with AF7 and formation of compact plasma fibrin clots resistant to lysis is observed in subjects with different types of AF.7,8 Impaired fibrinolysis has been reported to be associated with previous ischemic stroke in permanent AF.7 Treatment with vitamin K antagonists (VKA) alters clot structure by rendering fibrin clots more permeable and susceptible to fibrinolysis.9 Fibrin properties are modulated largely by environmental factors, including those associated with elevated risk for AF. Unfavorably altered fibrin structure, including dense fibrin networks, has been demonstrated in patients after ischemic stroke, with arterial hypertension, previous myocardial infarction (MI), type 2 diabetes mellitus, heart failure, and renal insufficiency.5 It is, however, unknown whether such prothrombotic clot phenotype increases thromboembolic risk in AF during follow-up.
Little is known about fibrin clot properties in bleeding disorders. It has been reported that increased fibrin clot permeability and susceptibility to lysis are associated with heavy menstrual bleeding of unknown cause,10 although unaltered clot lysability has also been demonstrated in this disorder.11 Hemophilia A and B are characterized by the formation of fibrin clots composed of thick fibers with increased lysability,12 and this mechanism has been suggested as a contributor to bleeding risk in this hemorrhagic diathesis.
Fibrin clot permeability, which indicates an average size of pores between fibrin fibers, is now the most commonly used and standardized hydrostatic pressure-driven assay to determine fibrin clot structure in various disease states.13,14 To our knowledge, there have been no reports on associations between the plasma clot phenotype and thrombotic or bleeding risk in AF. The aim of this longitudinal cohort study was to evaluate the relationship between fibrin clot density and outcomes among patients with AF treated with VKA.
Patients (n=296) with documented AF, treated with warfarin or acenocoumarol for at least 3 months, were recruited in outpatient clinics at the John Paul II Hospital in Cracow, Poland, between January 2010 and June 2012.
The ethical committee of the Jagiellonian University approved the study. All participants gave informed consent.
Data on demographics, cardiovascular risk factors, comorbidities, and drug therapy were collected prospectively, along with baseline clinical data. Ischemic stroke and a transient ischemic attack (TIA) were diagnosed based on the World Health Organization criteria. Labile international normalized ratio (INR) was defined as time in therapeutic range (TTR) of INR <65%. The CHA2DS2-VASc score was calculated to assess the risk for ischemic stroke and thromboembolism.15 The scores of 0, 1, and ≥2 were defined as the low, intermediate, and high-risk categories, respectively. The HAS-BLED score was used to evaluate the risk of bleeding in patients with AF.16 The scores of 0 to 2 and ≥3 were categorized as the low and high bleeding risk, respectively. Major bleeding was defined according to the International Society on Thrombosis and Hemostasis criteria as (1) fatal bleeding or (2) symptomatic bleeding in a critical area or organ and (3) bleeding accompanied by a decrease in the hemoglobin level of ≥2 g/dL or leading to transfusion of ≥2 U of whole blood/packed red blood cells. Minor bleeding was defined as any bleeding that did not meet the International Society on Thrombosis and Hemostasis criteria for major bleeding; occasional bruises <3 cm in diameter mostly on the extremities or self-declared occasional gingival bleeding were not included in this category.17 The exclusion criteria and other conditions were defined in the online-only Data Supplement.
Follow-up visits were performed on a 6-month basis (a visit at the center or telephone contact). Death, TIA, ischemic stroke, and major and minor bleeding were recorded and assessed by investigators unaware of the results of laboratory tests.
Laboratory investigations are available in the online-only Data Supplement.
Fibrin Clot Permeability
Fibrin clot permeability was determined using a pressure-driven system as described previously.14 Briefly, citrated plasma was mixed with 20-mmol/L calcium chloride and 1-U/mL human thrombin (Sigma-Aldrich, St Louis, MO). Tubes containing the clots were connected to a reservoir of a buffer (0.05 mol/L Tris–HCl, 0.15 mol/L NaCl, pH 7.5), and its volume flowing through the gels was measured within 60 minutes. A permeation coefficient (Ks), which indicates the pore size formed in fiber networks (lower values indicate tight fibrin structure), was calculated as follows: Ks=Q×Lη/t×A×Δp, where Q is the flow rate in time t, L is the length of a fibrin gel (13 mm), η is the viscosity of liquid (1/100 poise), A is the cross-sectional area (0.049 cm2), and Δp is a differential pressure (in dyne cm−2).13 The interassay and interassay coefficients of variation were <7%.
We studied 236 patients with AF (Table 1; Tables I through III in the online-only Data Supplement), including 125 (52.9%) subjects with permanent AF. Most patients (83.9%) were at high stroke risk (median CHA2DS2-VASc score, 3). A high bleeding risk (HAS-BLED score, ≥3) was found in 40.7% of patients. Patients were treated with acenocoumarol (58.1%) or warfarin (41.9%) for a median of 9 months (interquartile range, 5–35) before enrollment. A median TTR was 66.5% (interquartile range, 58–76). On the day of sampling 11.4% of patients had an INR value below 2.0, 17.8% had an INR value above 3.5, and 70.8% were within the range of INR between 2.0 and 3.5 and 62.2% within the range of 2.0 to 3.0.
Fibrin Clot Permeability
Ks ranged from 4.0 to 8.5 cm2×10−9. As expected, there was an inverse correlation between Ks and fibrinogen (r=−0.57; P<0.001) and a positive correlation with INR (r=0.31; P<0.001). Patients with TTR <65% had 5.8% lower Ks (6.5 [5.8–7.0] versus 6.9 [6.1–7.5] cm2×10−9; P=0.005). No associations of Ks with age, body mass index, type of VKA, comorbidities or medications, including aspirin, were observed in the patients with AF (Table II in the online-only Data Supplement). Multiple linear regression adjusted for potential confounders showed that fibrinogen (β=−0.58) and INR (β=−0.28) were the independent predictors of Ks (R2=0.38) in the entire AF group.
Patients with Ks below median (<6.8 cm2×10−9) had lower prevalence of permanent AF, higher TTR <65%, elevated fibrinogen, and lower INR (Tables 1 and 2). The proportions of patients with CHA2DS2-VASc scores of 0 to 1 and ≥2 and HAS-BLED scores 0 to 2 and ≥3 did not differ between the 2 groups.
Patients were followed for a total of 998 treatment-years with a median follow-up of 4.3 (3.7–4.8) years. None of the patients was lost to follow-up. Fifteen patients (6.36%) died. A mortality rate was 1.48% per year.
Ischemic Stroke or TIA
During follow-up, 30 stroke or TIA episodes were reported in 12.7% of patients (2.96% per year; Table 2). Patients with stroke or TIA had 10.8% (6.15 [5.5–6.5] versus 6.9% [6.0–7.4] cm2×10−9; P=0.0004) lower Ks measured at the enrollment compared with those free of cerebrovascular events (Table IV in the online-only Data Supplement).
The annual rate of these events was 6-fold higher in patients with low Ks (<6.8 cm2×10−9, median) versus the remainder (5.12% versus 0.79%; hazard ratio [HR], 6.55; 95% confidence interval [CI], 2.17–19.82; P=0.0009; Table 3; Figure 1). The independent predictors of stroke or TIA were Ks below 6.8 cm2×10−9 (β=−1.98; HR, 7.24; 95% CI, 2.53–20.76), age (β=0.04; HR, 1.05; 95% CI, 1.01–1.10), and use of angiotensin-converting enzyme inhibitors; (β=0.81; HR, 2.27; 95% CI, 1.08–4.77). The area under receiver operating characteristic curve of Ks for predicting stroke was 0.702 (95% CI, 0.610–0.793).
The optimal cutoff value of Ks for predicting stroke was 6.6 cm2×10−9. Sensitivity and specificity using this cutoff value were 86.7% and 59.2% (Figure 2).
Major and Minor Bleedings
During follow-up, 35 patients (14.8%) experienced a major bleeding (3.45% per year), including 16 patients (6.8%) with gastrointestinal bleeding (1.58% per year). The annual rate of intracranial bleedings was 0.39% (Table 2).
Patients with major bleeding and gastrointestinal bleeding had lower Ks (6.0 versus 6.9 cm2×10−9; P<0.0001 and 6.05 versus 6.9 cm2×10−9; P=0.008, respectively) measured at enrollment (Table IV in the online-only Data Supplement).
The annual rates of major bleedings was higher in patients with Ks <6.8 cm2×10−9 versus the remainder (6.11% versus 0.79%; HR, 10.65; 95% CI, 3.52–32.22; P<0.0001; Table 3; Figure 1). The annual rates of gastrointestinal bleedings were almost 15× higher in patients with low Ks (2.96% versus 0.20%; HR, 15.66; 95% CI, 1.95–126.1; P=0.01; Table 3). The independent predictors of major bleedings were Ks below 6.8 cm2×10−9 (β=2.13; HR, 8.48; 95% CI, 2.99–24.1) and HAS-BLED score of ≥3 (β=0.79; HR, 2.22; 95% CI, 1.12–4.38). Gastrointestinal bleeding was independently predicted by reduced Ks (β=2.75; HR, 15.07; 95% CI, 1.97–125.2) and aspirin (β=1.07; HR, 2.93; 95% CI, 1.08–7.49). The area under receiver operating characteristic curve of Ks was 0.755 (95% CI, 0.681–0.829) for predicting major bleeds and 0.751 (95% CI, 0.667–0.834) for gastrointestinal bleeding. The optimal cutoffs for Ks were 6.5 cm2×10−9 (sensitivity 82.9% and specificity 62.7%) and 6.7 cm2×10−9 (sensitivity 93.8% and specificity 53.2%), respectively (Figure 2).
During follow-up, 77 (32.6%) patients reported minor bleedings. Minor bleeds were reported in 11.63% of patients with Ks above 6.8 cm2×10−9 and in 3.55% of those with Ks below median (odds ratio, 5.23; 95% CI, 2.60–10.53; P<0.0001). Patients with minor bleeds had 15.6% (7.4 [6.9–7.8] versus 6.4 [5.8–6.9] cm2×10−9) higher Ks measured at the enrollment compared with those without such bleeding (P<0.0001). The area under receiver operating characteristic curve of Ks for predicting minor bleeding was 0.774 (95% CI, 0.711–0.774). The Ks value of 7.0 cm2×10−9 offered the best overall sensitivity and specificity, namely 72.7% and 76.1%, respectively (Figure 2).
Ks below 6.8 cm2×10−9 was associated with 10-fold higher risk of composite end point involving stroke or TIA and major bleeding (HR, 10.21; 95% CI, 4.43–23.53; P<0.00 001). Ks did not show any association with all-cause mortality (Figure I in the online-only Data Supplement; Table V in the online-only Data Supplement).
Our study indicates that fibrin structure has a prognostic value in patients with AF already being treated with VKA. This study is the first to show that more compact fiber networks, reflected by lower plasma clot permeability measured ex vivo, that is, Ks, are independently associated with an increased risk of ischemic stroke or TIA and major bleeding in patients with AF treated with VKA. Moreover, looser fibrin networks observed during anticoagulation with VKA predisposed to mucocutaneous bleedings during follow-up. Our findings provide evidence that unfavorable clot structure measured on VKA may independently affect both thromboembolic and bleeding risk. Our study indicates that fibrin structure reflected by Ks provides additional prognostic information about stroke and major bleeding risks in AF that cannot be extracted from clinical scores. Based on the Ks value, it is possible to identify additional anticoagulated patients with AF at risk of adverse clinical outcomes.
The prevalence of stroke and major bleeds in our cohort is similar to that reported in other studies.19 Clinical characteristics in this study were representative to typical AF patients.15,20 Similarly to other studies, the HAS-BLED score predicted major bleeding with moderate accuracy.20 Although low TTR is considered a risk factor of death, myocardial infarction, major bleeding, and stroke, its effect was diminished because of stable anticoagulation in our study.21
Fibrin permeability is determined predominantly by plasmatic and platelet-derived proteins involved in hemostasis; therefore, the impact of clinical risk factors on fibrin clot characteristics is of minor importance as was demonstrated previously.22
Fibrin clot permeability, Ks as a measure of fibrin architecture, has been evaluated in various prothrombotic disorders using a few plasma-based assays in which clotting is triggered by tissue factor or thrombin.13 Recently, we have developed a novel automated method for determining fibrin porosity, which shortens the measurement time.23
It is known that decreased Ks is associated with arterial thromboembolic complications of atherosclerotic vascular disease.5 However, we provided the first evidence that compact fibrin networks determined on the maintenance VKA therapy has a negative impact on prognosis in patients with AF, suggesting that despite fluctuations in INR values during follow-up, lower Ks predisposes to symptomatic thromboembolic events. The most likely mechanism underlying the increased risk of stroke in subjects who form denser fibrin meshwork is impaired efficiency of plasmin-mediated lysis of such structures as shown previously.18 Importantly, we demonstrated that denser fibrin networks can be observed in anticoagulated patients. This suggests that VKAs do not abolish unfavorable clot characteristics typical of patients with AF.9 In our patients, a beneficial effect of VKA was partially abolished by cardiovascular risk factors and concomitant diseases. It has been shown that treatment with VKA results in formation of permeable clots with a mean increase in Ks by ≈16%.9 It should be noted that in the present study, Ks is weakly influenced by INR in the TTR, which indicates that the structure of fibrin clots is governed by fibrinogen levels and quality, along with other factors, including von Willebrand factor (VWF), platelet factor 4,8 that do not affect INR values used mainly to monitor VKA therapy.
It is known that VKAs modify fibrin clot properties early with plateau after 5 to 7 days; however, they have limited effect on fibrin network compared with various disease states.9,22 It is probable that VKAs improve clot phenotype in patients with AF to a varying extent, and the magnitude of changes in Ks assessed during anticoagulation is determined by genetic and environmental factors affecting individual features of a fibrin clot in a given patient, that is, activity of vitamin K-dependent coagulation factors, especially factor IX as reported previously.9 The regulation of VKA-induced alterations to fibrin clots in patients with AF remains to be established.
Our unexpected finding is that dense fibrin networks are also associated with major bleedings during VKA therapy. Mechanisms behind complex interactions between fibrin permeability and bleeding remain to be established. Fibrin clot structure is an important determinant of wound healing affecting cell adhesion, migration, proliferation, and vessel formation.24 Disturbances in wound healing are observed in patients with unfavorably modified fibrin porosity.25 Bleeding risk may be related to function of other proteins determining Ks, for example, tPA (tissue-type plasminogen activator) or VWF. Roldán et al26 have reported that patients with AF with high levels of VWF—a risk factor of thromboembolic complications—that is associated with dense fibrin clots27 had an increased risk of bleeding complications during VKA treatment.19 Moreover, increased P-selectin and platelet factor 4 observed in AF have also been shown to reduce Ks in this disease.8,28 As expected, minor bleedings were more common in ssubjects with looser fibrin architecture, as evidenced by higher Ks, which agrees with observations in inherited bleeding disorders.10,12 Our findings suggest that less-dense fibrin architecture facilitating efficient fibrinolysis and disintegration of forming hemostatic plugs is of importance mainly after mild skin injury or the mucous membrane in the oral cavity. Our study supports the view based on molecular analysis that more permeable clots enhance susceptibility to lysis by facilitating tPA and plasminogen binding to fibrin.5 In patients with AF on VKA, Ks reflects effects of several (poorly characterized yet) factors affecting the efficiency of blood coagulation and hemostasis that are implicated in serious complications of the VKA use in AF and possibly in other disorders. The quality of fibrin fiber networks, their density implicating mechanical characteristics, for example, elasticity and lysability, are of key importance for hemostasis, thrombosis, and bleeding; we have reported here that they are important also in anticoagulation-related bleeding.
Interestingly, we found that angiotensin-converting enzyme inhibitors, which were prescribed largely to subjects with arterial hypertension or after MI, may predict stroke or TIA, but it rather reflects more comorbidities and treatment of patients at high thromboembolic risk. In stable coronary artery disease, angiotensin-converting enzyme inhibitors have been demonstrated to increase plasma clot permeability that might contribute to reduction in thrombotic complications.29 It might be speculated that these agents are not potent enough to increase Ks in patients with AF, similarly to no or negligible effects of other potential favorable modulators of fibrin clots, such as statins and aspirin.5,6
Based on the current evidence, the decision to start anticoagulation should be based on the stroke risk assessment. Even high bleeding risk is not considered the contraindication to initiate anticoagulation. Identification of AF patients with low Ks can lead to closer surveillance during VKA therapy and more frequent visits in outpatient clinics.
It might be proposed that assessment of clot permeability can optimize the management of anticoagulated patients with AF. Left atrial appendage closure remains an alternative for stroke prevention in patients with an elevated stroke risk and high risk of anticoagulation failure or hemorrhagic complications. Fibrin clot permeability testing may help to identify patients with the highest risk of ischemic stroke or TIA and major bleeding who potentially benefit most from left atrial appendage closure. However, this hypothesis requires further studies.
There are several study limitations. First, plasma blood samples were collected only once during VKA treatment; therefore, we cannot assess the magnitude of changes induced by VKA. A size of the study was limited; however, the number of subjects was sufficiently powered, and we found similar complication rates in our patients as in the AF registries.19 Statistical associations reported here do not necessarily mean a cause–effect relationship. We did not investigate biomarkers that are associated with bleeding and thromboembolic risk in AF to compare their predictive value with Ks. It is also unclear whether in AF ex vivo plasma clot structure correlates with the amount and architecture of fibrin component in thrombi forming in AF; however, Zalewski et al4 showed such associations in patients with acute MI in whom intracoronary thrombi were retrieved during thrombectomy. Finally, it remains to be established whether therapy with nonvitamin K oral anticoagulants improves clot phenotype like VKA in AF with a similar prognostic potential.30
Our findings demonstrate that abnormal structure of fibrin network in AF contributes to thrombotic and bleeding complications during treatment with VKA. Further studies are needed to validate our observations.
Sources of Funding
This study has been supported by a grant from Jagiellonian University Medical College (K/ZDS/005802 to Dr Undas).
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.117.018143/-/DC1.
- Received May 21, 2017.
- Revision received August 12, 2017.
- Accepted August 15, 2017.
- © 2017 American Heart Association, Inc.
- Kirchhof P,
- Fabritz L
- Zalewski J,
- Bogaert J,
- Sadowski M,
- Woznicka O,
- Doulaptsis K,
- Ntoumpanaki M,
- et al
- Drabik L,
- Wołkow P,
- Undas A
- Ząbczyk M,
- Majewski J,
- Karkowski G,
- Malinowski KP,
- Undas A
- Szczepaniak P,
- Zabczyk M,
- Undas A
- Wiewel-Verschueren S,
- Knol HM,
- Lisman T,
- Bogchelman DH,
- Kluin-Nelemans JC,
- van der Zee AGJ,
- et al
- Undas A
- Pieters M,
- Undas A,
- Marchi R,
- De Maat MPM,
- Weisel JW,
- Ariens RAS
- Olesen JB,
- Lip GYH,
- Hansen ML,
- Hansen PR,
- Tolstrup JS,
- Lindhardsen J,
- et al
- Schulman S,
- Kearon C
- Roldán V,
- Marín F,
- Muiña B,
- Torregrosa JM,
- Hernández-Romero D,
- Valdés M,
- et al
- Lip GYH,
- Frison L,
- Halperin JL,
- Lane DA
- Kornej J,
- Potpara T,
- Lip GYH
- Undas A,
- Ariëns RAS
- Ząbczyk M,
- Piłat A,
- Awsiuk M,
- Undas A
- Wannamethee SG,
- Whincup PH,
- Lennon L,
- Rumley A,
- Lowe GD,
- Lau YC,
- Xiong Q,
- Shantsila E,
- Lip GYH,
- Blann AD