Plasminogen activation and plasmin activity are not necessary to prevent venous thrombosis/thromboembolism Open Access

Venous thrombosis without or with pulmonary embolism (collectively venous thromboembolism [VTE]) affects 1 to 2 per 1000 individuals annually and is a major cause of morbidity and mortality.¹ Venous thrombosis is initiated by exuberant or dysregulated thrombin generation that promotes intravascular fibrin deposition, which traps resident cells and produces an occlusive clot. Fibrin also promotes tissue plasminogen activator (tPA)–mediated conversion of plasminogen to plasmin, which can cleave fibrin and facilitate clot degradation. Enhanced plasmin(ogen) activation/activity (PA) can increase bleeding risk in humans.^2,3 Conversely, suppression of PA with antifibrinolytic agents (eg, tranexamic acid [TXA]) reduces bleeding and decreases hemorrhage-related deaths in patients with surgical or trauma-induced bleeding and in women with postpartum hemorrhage.^4-10 TXA also enhances fibrin accumulation in a laser injury-induced mouse model of hemostasis.¹¹ Extension of these observations has led to the concern that reduced or suppressed PA may facilitate intravascular fibrin formation and increase VTE risk. Accordingly, antifibrinolytic therapy may be withheld from patients with increased thrombosis risk or active intravascular clotting, and these patients are often excluded from trials of TXA.⁴ 

The premise that PA is needed to prevent intravascular fibrin accumulation is supported by in vitro studies implicating PA pathway-related proteins in VTE risk. Studies of plasma-based clot lysis assays associated a prolonged clot lysis time with increased risk of first VTE,¹² and both lysis time and VTE risk were positively associated with levels of the antifibrinolytic proteins plasminogen activator inhibitor-1 (PAI-1) and thrombin activatable fibrinolysis inhibitor (TAFI).¹³ Additionally, several clinical trials assessing PA suppression with TXA in individuals with severe injury, acute gastrointestinal bleeding, or heavy menstrual bleeding associated TXA use with increased VTE.^14-19 However, limitations of these studies include potential survivorship bias, likelihood that more severely injured patients received TXA, presence of comorbid disease, and relatively minor overall risk.^4,14-20 

Other large trials of all-cause trauma, traumatic brain injury, postpartum hemorrhage, intracerebral hemorrhage, and coronary artery surgery have not detected increased VTE following TXA treatment.^5-8,21 Furthermore, systematic reviews and meta-analyses have not associated TXA with increased risk of arterial thrombosis or VTE in bleeding, surgical, or nonsurgical patients.^22-25 Although sporadic studies identified VTE in plasminogen-deficient individuals,^26-30 larger studies did not,^31,32 and VTE has not been widely reported in these individuals.^33-36 However, congenital plasminogen deficiency is ultrarare (∼1.6 per million individuals), and historic population studies have lacked sufficient real-world data to assess rare complications.

Here, we studied mice and humans using genetic, pharmacologic, and proteomic analytical strategies to define the impact of PA on VTE. Our data indicate that loss or suppression of PA does not increase venous thrombus formation and suggest PA is not a routine biological surveillance mechanism to prevent VTE.

The Hypoplasminogenemia International Retrospective and Prospective Cohort Study (HISTORY) protocol was approved by the Ascension Health Institutional Review Board (IRB; protocol number R20170057; ClinicalTrials.gov #NCT03797495, https://clinicaltrials.gov/ct2/show/NCT03797495). The use of human participants was also approved by the local IRB or Ethical Committee for each participating site before enrolling study participants. Each participant provided IRB/Ethical Committee–approved written informed consent. Proband and first-degree relatives were enrolled in this 4-year international, multicenter retrospective/prospective registry.^33,37 The study design is described in the supplemental Materials (available on the Blood website).

Murine studies were approved by the University of North Carolina at Chapel Hill (UNC) Institutional Animal Care and Use Committee (protocol number 22-019). Genetically engineered mice on a C57BL/6J background that do or do not express plasminogen (Plg^+/+, Plg^+/−, and Plg^−/−) were described previously.³⁸ Wild-type C57BL/6J mice from the Jackson Laboratories (Raleigh, NC) were maintained by homozygous breeding. Female and male mice (8-12 weeks old) were used for in vitro and ex vivo experiments. Only male mice were used for the infrarenal inferior vena cava (IVC) ligation venous thrombosis model, as ligation of lateral IVC branches can result in reproductive organ necrosis in female mice.³⁹ Mice were anesthetized with 1.5% isoflurane in oxygen (2 L/min) or ketamine/xylazine (100/15 mg/kg).

Mouse blood was collected, and platelet-poor plasma (PPP) was prepared as detailed in the supplemental Materials. To quantify TXA effects in vitro, TXA (catalog number PHR1812, Sigma, Burlington, MA) was prepared in 20 mM HEPES, 150 mM NaCl (HBS), and added to normal pooled mouse plasma.

Plasma PA was measured as described⁴⁰ and detailed in the supplemental Materials.

The venous thrombosis ligation model was performed as described⁴¹ and detailed in the supplemental Materials.

Frozen thrombi were embedded in optimal cutting temperature compound, sectioned longitudinally and consecutively (5-μm sections), and mounted on slides by serial interruption (UNC Histology Research Core). Slides were treated to identify nuclei/DNA, platelets, and fibrin(ogen) using DAPI (4′,6-diamidino-2-phenylindole), anti-CD41, and anti-fibrin(ogen), respectively. Images were analyzed using Fiji (ImageJ 1.54f, National Institutes of Health). Detailed methods for immunofluorescence and image analysis are provided in the supplemental Materials.

Mice were administered TXA (600 mg/kg⁴² at 20 μL/g body weight) or saline by intraperitoneal injection. Blood was collected from the IVC of anesthetized mice into 3.2% sodium citrate (10% vol/vol, final) at the indicated times after injection. PPP was prepared, aliquoted, flash frozen, and stored at −80°C.

TXA concentrations in plasma were measured by mass spectrometry (UNC Mass Spectrometry Core Laboratory). Details are provided in the supplemental Materials.

Alkaline phosphatase, alanine transaminase, aspartate transaminase, blood urea nitrogen, and creatinine were measured in plasma using a Vet Axcel Clinical Chemistry System (UNC Animal Clinical Chemistry Laboratory).

Fibrin degradation products (FDPs) were measured using an enzyme-linked immunosorbent assay (catalog number CSB-E07941, Cusabio, Houston, TX) per the manufacturer’s instructions. Plasma from Fga^–/– mice was used to quantify nonspecific background signal, which was then subtracted from all measurements after calculating FDP concentrations; values <0 were assigned as 0.

Analyses were performed with GraphPad Prism version 10.4.0. Data normality was assessed by Shapiro-Wilk tests. Statistical tests used for group comparisons are indicated in the figure legends. P < .05 was considered significant unless otherwise indicated.

Summary data from a cross-ancestry meta-analysis genome-wide association study (GWAS) of VTE by the International Network Against Venous Thrombosis Consortium⁴³ were used to examine variants in 3 groups of genes (positive control, negative control, and hypothesis-based PA pathway genes) as described in the supplemental Materials. We first visually compared the distribution of P values observed for each gene with the distribution expected with a null distribution. We further tested variants for association with VTE using a Bonferroni-corrected threshold for significance that was set at P < 3.4 × 10^–6 (.05/14 688 variants tested).

Associations of 7 hypothesis-based PA pathway-related proteins (α₂-antiplasmin, TAFI, plasminogen, urokinase plasminogen activator [uPA], tPA, PAI-1, and [soluble] uPA receptor [uPAR]) with incident VTE were assessed using Olink protein profiles data from the UK Biobank,^44,45 as described in the supplemental Materials. The Bonferroni-corrected threshold for significance was set at P < .007 (.05/7 proteins tested).

VTE has been observed in individuals with plasminogen deficiency; however, these reports have primarily been case reports and small family studies.^26-30 To more rigorously characterize VTE risk in humans with congenital plasminogen deficiency, we reviewed data from the HISTORY registry, an international cohort of individuals with congenital type 1 plasminogen deficiency and their first-degree relatives. The recently analyzed cohort with complete data includes 107 participants: 46 (43%) homozygous/compound heterozygous-deficient, 49 (46%) heterozygous-deficient, and 12 (11%) wild-type participants. Of the 46 homozygous/compound heterozygous-deficient individuals, 28 (61%) were female, the median age at entry was 15.1 years (range, 2.5-75.9 years), the median plasminogen activity was 15% (range, 1%-44%; normal range, 80%-132%), and median plasminogen antigen was 20.5 μg/mL (range, 0.0-68.6 μg/mL; normal range, 70-215 μg/mL). Of the 49 heterozygous-deficient individuals, 26 (53%) were female, the median age at entry was 41 years (range, 4.2-71.4 years), the median plasminogen activity was 59% (range, 41%-84%), and the median plasminogen antigen was 79.9 μg/mL (range, 46.3-143.2 μg/mL). Control participants had plasminogen activity and antigen levels within the normal range.

As published,⁴⁶ the most common symptoms reported in homozygous/compound heterozygous individuals were ligneous conjunctivitis and gingivitis; decreased and variable frequency of symptoms in other systems including ear, oropharyngeal, respiratory, gastrointestinal, genitourinary, and central nervous system were also reported. Notably, however, no individuals with heterozygous, homozygous, or compound heterozygous deficiency reported a history of venous thrombosis or pulmonary embolism.

Because VTE may be rare even in a relatively large registry of plasminogen-deficient humans, we then tested the role of PA in venous thrombogenesis using mice with genetic loss of plasminogen (Plg^+/−, Plg^−/−). Similar to plasminogen-deficient humans, previous studies documented ligneous conjunctivitis and spontaneous fibrin deposition in liver, stomach, and other organs of Plg^−/− mice, but have not identified spontaneous thrombosis in large veins.^38,47,48 As expected,⁴⁰ compared with Plg^+/+ mice, PA in PPP from Plg^+/− and Plg^−/− mice was partially reduced or completely absent, respectively (Figure 1A). However, when subjected to an experimental model of venous thrombosis,⁴¹ Plg^+/+, Plg^+/−, and Plg^−/− mice produced thrombi with similar mass 24 hours after thrombus induction (Figure 1B). Because venous thrombi reach peak mass in this model within 24 to 48 hours,³⁹ we speculated that it may be difficult to detect thrombus-enhancing effects at this time point. Therefore, to determine if loss of plasminogen increases thrombus mass earlier during thrombogenesis, we also analyzed thrombi harvested after 6 hours (Figure 1C). As expected, thrombi harvested at 6 hours were smaller than thrombi at 24 hours (Figure 1B-C). However, the mass of thrombi from Plg^+/− and Plg^−/− mice did not differ from Plg^+/+ mice, even at this early timepoint (Figure 1C).

Because the loss of PA may increase fibrin(ogen) stability and, therefore, enhance fibrin(ogen)-dependent recruitment of leukocytes and/or platelets to the thrombus, we used immunofluorescence to visualize thrombus components 6 hours after thrombus induction. Quantification of these images showed similar nuclear, platelet, and fibrin(ogen) content in thrombi from Plg^+/+, Plg^+/−, and Plg^−/− mice (Figure 1D-E). Collectively, these data show that complete loss of PA does not increase venous thrombus mass or alter thrombus composition in mice.

TXA inhibits tPA-mediated PA in human PPP.^4,49 To characterize the effect of pharmacologically suppressing PA in mice, we first confirmed the sensitivity of the PA assay to TXA in mouse plasma by adding a therapeutic concentration range of TXA (0.0-60.8 μg/mL, final) to normal pooled mouse plasma and measuring PA kinetics. As expected, TXA dose-dependently delayed the lag time and time to peak, and suppressed the peak and velocity, and 60.8 μg/mL TXA completely inhibited PA (Figure 2A; supplemental Figure 1). To characterize the pharmacodynamic effect of TXA in mice in vivo, we infused TXA (600 mg/kg) via intraperitoneal injection, collected blood, and measured plasma PA potential as a function of time after TXA administration. As expected, administration of saline (vehicle control) did not alter PA; however, TXA administration delayed the lag time and time to peak, and decreased the velocity and peak (Figure 2B-G). Effects of TXA on PA were lost within 3 hours after administration, consistent with its short half-life in humans.^50,51 These data show TXA has similar pharmacokinetic and pharmacodynamic properties in healthy mice as in humans.

To then determine the effect of pharmacologically suppressing PA on venous thrombus formation in mice, we triggered thrombosis by IVC ligation in wild-type mice treated with TXA or saline control (first dose immediately after IVC ligation, second dose 12 hours after IVC ligation; Figure 3A). As seen in plasminogen-deficient mice, the mass of thrombi harvested from TXA-treated wild-type mice did not differ from saline-treated mice at either 24 or 6 hours after thrombus induction (Figure 3A-B). Likewise, thrombi harvested from TXA-treated mice did not differ from saline-treated mice in nuclear, platelet, or fibrin(ogen) content at either time point (Figure 3C-F). These data suggest pharmacologic suppression of PA does not increase venous thrombus mass or alter thrombus cellular or fibrin content.

Given the short half-life of TXA in both humans^50,51 and mice (Figure 2), we also tested whether the lack of effect of TXA on venous thrombosis was due to its rapid clearance. We measured circulating TXA in mice subjected to IVC ligation using functional (PA assay) and quantitative (mass spectrometry) methods. As expected, PPP collected from saline-treated mice subjected to IVC thrombosis model showed detectable PA (Figure 4A-B). Unexpectedly, and in contrast to TXA-treated mice that were not subjected to IVC ligation surgery (Figure 2), PPP collected from several of the TXA-treated mice subjected to IVC ligation exhibited substantially reduced PA even up to 12 hours after the last dose of TXA (Figure 4A-B). Mass spectrometry revealed residual TXA in these plasmas (Figure 4C), which correlated with the plasmin peak (Spearman r = −0.87, P < .001; Figure 4D). TXA undergoes renal clearance in humans.⁴ Interestingly, both creatinine and blood urea nitrogen—markers of kidney function—as well as alanine transaminase and aspartate transaminase—markers of liver function—were elevated in both saline- and TXA-treated mice subjected to 24 hours of IVC ligation (supplemental Figure 2). These data suggest that the surgery and/or IVC ligation compromise kidney and liver function and alter TXA metabolism (absorption and/or clearance). Nonetheless, because TXA was preserved in the plasma of treated mice, we then measured FDP in PPP 6 and 24 hours after ligation. Interestingly, elevated levels of FDP were not detected in saline-treated mice until 24 hours after thrombus induction, suggesting little-to-no fibrinolysis occurred during early thrombogenesis (Figure 4E). At 24 hours, FDPs were elevated in saline-treated mice but not in TXA-treated mice (Figure 4E) but still represented <3% of total estimated fibrin(ogen) in a thrombus (60 ± 34 μg fibrin(ogen)/13 ± 3 mg thrombus, data not shown). Collectively, these findings suggest genetic or pharmacologic loss of PA does not increase thrombus mass because little-to-no fibrinolysis occurs within the first 24 hours of venous thrombus formation.

We then expanded our investigation to a genomic analysis of VTE risk in a large human cohort. Using summary data from the largest VTE GWAS to date,⁴³ we extracted P values for variants 50 kilobases upstream and downstream of selected PA pathway-related genes, as well as positive and negative control genes. Across all genes in the 3 groups, we identified 14 688 variants, which were primarily single-nucleotide substitutions. As expected, P-value distributions for 5 (anti)coagulation genes (F5, FGG, F2, PROC, and SERPINC1) and VTE were smaller than the expected null distribution (Figure 5). In contrast, P-value distributions for genes selected for a lack of prior association with VTE (negative controls) as well as genes encoding proteins with roles in the PA pathway (hypothesis-based genes) did not differ from the expected null distribution (Figure 5). On testing, all positive control genes had many (between 24 and 279) variants that exceeded the threshold for statistical significance (0.05/14 688 = 3.4 × 10^–6), whereas the negative control and hypothesis-based gene groups each had only 1 gene (TGM5 and SERPINF2, respectively) with single variants that minimally exceeded the threshold for significance (Table). These data suggest VTE risk is not substantially altered by genetically determined modifications in PA pathway genes.

Finally, we analyzed proteomic data from the UK Biobank^44,52 to characterize relationships between VTE and 7 PA pathway-related proteins. As expected,⁵³ the plasma level of (soluble) uPAR was significantly associated with VTE risk (P < .001), and this association persisted even after further adjustment for biomarkers of chronic inflammation (C-reactive protein or fibrinogen; Figure 6; supplemental Table). Consistent with a previous case-control study,¹³ both PAI-1 and tPA were also significantly associated with VTE (Figure 6; supplemental Table). Further adjustment for body mass index, a biomarker for obesity that enhances both PAI-1⁵⁴ and VTE risk,⁵⁵ did not remove the significant association of PAI-1 with VTE; however, further adjustment for body mass index and PAI-1, which contributes to plasma tPA levels,⁵⁶ diminished the association between tPA and VTE (Figure 6; supplemental Table). Importantly, however, neither plasminogen, α₂-antiplasmin, TAFI, nor uPA levels were significantly associated with VTE risk (Figure 6; supplemental Table).

Collectively, these data from clinical observations of humans with congenital type 1 plasminogen deficiency, experimental genetic and pharmacologic mouse models of PA loss/suppression, a large GWAS for VTE risk, and plasma protein levels in humans indicate that PA is not a biological surveillance mechanism to protect against VTE.

Genetic or pharmacologic loss of PA can promote extravascular fibrin formation,^38,47,48 reduce bleeding, and decrease hemorrhage-related deaths in patients with surgical or trauma-induced bleeding.^4-10 However, whether the endogenous PA system serves as a biological mechanism to prevent intravascular thrombus formation has been debated. Accordingly, concerns that therapeutic suppression of PA may increase VTE risk have limited the use of antifibrinolytic therapies. We used a multidisciplinary approach to directly test the hypothesis that loss of PA increases VTE. Our findings from each of these methods consistently show that loss of or common perturbations in PA genes or proteins does not enhance VTE risk. Collectively, these findings suggest PA has little-to-no role in preventing thrombus formation, inform future studies using experimental models to define VTE pathophysiology, and suggest that antifibrinolytic therapy may be used to staunch bleeding without increasing VTE risk.

Despite the therapeutic efficacy of enhanced PA to dissolve existing thrombi^57,58 and prevent thrombus formation,^59,60 the present findings from mice and humans suggest the inverse is not true; impaired ability to generate plasmin does not increase VTE risk. Although conceptually paradoxical, these data are consistent with prior observations. First, humans with type 1 plasminogen deficiency often experience inflamed growths on mucous membranes, including the eyelids and mouth, but VTE is uncommon.^35,36 Similarly, Plg^−/− mice develop ligneous conjunctivitis and extravascular fibrin deposition, but do not spontaneously produce occlusive thrombi within large vessels,^38,47,48 and mice genetically engineered to recapitulate the type 2 plasminogen Tochigi variant have phenotypes similar to wild-type mice even after prothrombotic challenges.⁶¹ Second, like Plg^−/− mice, mice with deficiencies in proteins that activate plasminogen (tPA, uPA) also do not have larger thrombi 24 hours after IVC ligation.⁶² Third, repeated TXA treatment in women with renal dysfunction and impaired TXA clearance induces an acquired hypoplasminogenemia and ligneous conjunctivitis but not VTE.^63,64 Fourth, meta-analyses of trials assessing TXA use for bleeding in surgical or nonsurgical patients have not detected increased VTE risk.^22-25 Fifth, although GWASs have significantly associated numerous genes, including several encoding proteins with procoagulant (eg, factor V, prothrombin) or anticoagulant (eg, protein C, antithrombin) functions with VTE risk,⁴³ this has not been the case for genes encoding proteins within the PA pathway. Sixth, although case-control studies associated delayed clot lysis in vitro with increased risk of primary VTE, risk was primarily driven by PAI-1, and an unexpected positive association of plasminogen with VTE was thought to reflect altered levels of other coagulation and fibrinolytic factors.^12,13 Finally, a review of studies evaluating fibrinolytic activity and VTE did not identify a significant association between impaired fibrinolysis and increased VTE.⁶⁵ Collectively, the substantial concordance of the present findings with these reports suggests that despite its well-recognized ability to cleave fibrin, loss of PA does not enhance VTE risk.

Because VTE is relatively uncommon in adolescence and early adulthood,¹ VTE is unlikely to have driven the strong evolutionary conservation of PA pathway proteins and biochemical functions.^66,67 Finding that PA is not biologically tasked with venous thrombus prevention suggests its major role is in mitigating extravascular, rather than intravascular, fibrin deposition, as well as other essential processes, including wound healing, inflammation, and innate immunity.^67,68 For example, plasminogen mediates inflammation, proliferation, and tissue remodeling during acute wound healing.⁶⁹ Plasminogen also exerts beneficial effects during infections but can also exacerbate sepsis by altering proinflammatory and anti-inflammatory cytokine profiles.⁷⁰ The complex roles of PA pathway proteins are further underscored by the associations of (soluble) uPAR and PAI-1 but not plasminogen with VTE, indicating PA-independent roles for these proteins. uPAR is a cell surface receptor and its shedding into plasma is associated with acute and chronic inflammation. In addition to its role in PA, uPAR is also involved in cell signaling⁷¹ and leukocyte recruitment,⁷² each of which has been implicated in VTE pathogenesis. Likewise, in addition to inhibiting tPA and uPA, PAI-1 interacts with other proteins implicated in VTE (eg, FXI[a])⁷³ and has been linked to prothrombotic pathologies, including inflammation, metabolic syndrome, and fibrosis.⁷⁴ Adjusting for potential contributions of all VTE risk factors is beyond the scope of the present study but should fuel future investigations.

An unexpected observation from this study was the prolonged time with which TXA circulated and inhibited plasma PA in mice subjected to IVC ligation. The infrarenal IVC ligation model is frequently used for venous thrombosis studies because this model recapitulates the slow formation of large, occlusive thrombi that are enriched in fibrin and red blood cells.^39,75 Accordingly, IVC ligation models have been used to elucidate fundamental (patho)physiologic mechanisms that promote thrombus formation and test the antithrombotic activity of novel pharmacologic agents.^76-78 The observation that both TXA, which is eliminated by the kidneys in humans,⁴ as well as markers of kidney and liver dysfunction are elevated in mice subjected to IVC ligation suggests the surgery and/or ligation procedure compromises functions associated with these organs. Thus, the IVC ligation model, itself, may alter absorption and/or clearance of therapeutics and coagulation-related proteins, which may overweight or underweight endogenous mechanisms mediating venous thrombosis or the efficacy of certain therapeutics. Although speculative, this limitation may impede translation from the preclinical to clinical setting and warrants continued study.

A major strength of our study was the comprehensive, multidisciplinary approach that used data from: (1) a large international study of humans with congenital type 1 plasminogen deficiency, (2) mouse models of both genetic and pharmacologic suppression of PA in an established experimental model of venous thrombosis,⁴¹ (3) a large GWAS for VTE risk alleles, and (4) a large prospective proteomics study. This comprehensive, orthogonal approach compensated for potential limitations of each method and leveraged the strengths of each: Plasminogen deficiency in humans is ultrarare and even a large registry of these individuals may not capture rare events; however, the lack of reported VTE in this international retrospective/prospective registry is consistent with meta-analyses of large clinical trials that failed to associate pharmacologic inhibition of plasmin activity with VTE.^22-25 Although mice only partly recapitulate PA pathway biochemistry in humans,⁷⁹ they allow for both genetic and pharmacologic manipulation of the PA system. Moreover, the mouse IVC ligation model enables rigorous characterization of thrombogenesis and thrombus composition. Although it is unclear how much TXA reached the IVC following ligation, similar findings in Plg^−/− mice that lack the capacity to generate plasmin reinforces the premise that loss of plasmin potential does not enhance thrombogenesis in mice. The International Network Against Venous Thrombosis summary data did not include rare variants (mean allele frequency <0.01); however, it is our experience that it is uncommon to find rare variants associated with VTE in the absence of common variant associations. Although genomic analyses do not capture the contribution of abnormal protein levels to outcomes, data from the UK Biobank also suggest plasminogen levels are not prospectively associated with VTE risk. Notably, concordance in findings across these approaches refutes the hypothesis that PA deficiency predisposes individuals to VTE.

Important questions remain. First, it is unclear whether these findings can be extrapolated to the use of antifibrinolytic therapy in all clinical populations. Although both our data and meta-analyses of large clinical trials suggest TXA does not significantly increase VTE incidence,^22-25 potential effects of TXA on other aspects of VTE or on arterial thrombosis remain unclear. For example, TXA inhibits tPA-mediated PA but promotes uPA-mediated PA,⁸⁰ and plasmin has nonfibrin substrates in blood (eg, complement C5, factors IX[a], V, and VIII) and on the surface of cells.^67,81-85 Thus, changes in PA dynamics may facilitate or suppress downstream signaling pathways and/or inflammatory responses. Second, our study focused on thrombus formation, but did not address thrombus resolution. PAI-1– and PAI-2–deficient mice have enhanced thrombus resolution,^86,87 whereas uPA-deficient mice have impaired monocyte influx and delayed thrombus resolution.⁶² Thus, although reduced PA does not appear to increase venous thrombogenesis, plasminogen deficiency may delay resolution and/or increase risk of postthrombotic sequelae. Third, the presence of additional pathophysiologic insults may enhance the need for PA and should be explicitly tested in future studies. For example, robust PA may be needed to prevent VTE in settings in which normal coagulation mechanisms become substantially dysregulated (eg, extensive trauma/tissue injury,⁸⁸ malignancy, and patients with catheters or other artificial surfaces). Further work is needed to define the role of PA in these complicated settings.

In summary, our data show that neither genetic nor pharmacologic loss or perturbations of PA increase VTE risk. These data indicate PA does not protect against thrombus formation and suggest that in routine situations, antifibrinolytic agents may be used to prevent bleeding without increasing VTE risk.

The authors thank Pablo Ariel at the UNC Microscopy Service Laboratory for help with immunofluorescence imaging and analysis, Brandie M. Ehrmann and Cristina Arciniega at the UNC Department of Chemistry Mass Spectrometry Core Laboratory for help with TXA mass spectrometry analysis, the UNC Animal Clinical Chemistry Services Core, Keely Glass for technical support with the venous thrombosis model in plasminogen-deficient mice, Samer M. Issa for assistance with mouse experiments, Bryce Kerlin for helpful discussions, and Stéphanie E. Reitsma for reading the manuscript.

This work was supported by funding from the American Heart Association (23POST1017878 to Y.S.) and the National Institutes of Health (NIH), National Heart, Lung, and Blood Institute (HL126974 [A.S.W.], HL147894 [N.L.S., A.S.W.], HL143403 [M.J.F., A.S.W.], HL168009 [M.J.F.], and HL139553 and HL154385 [N.L.S.]). This work was also partially supported by the Italian Ministry of Health–Bando Ricerca Corrente (F.P.). The Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico is a member of the European Reference Network (ERN) on Rare Haematological Diseases EuroBloodNet-Project number 101157011. ERN-EuroBloodNet is partly cofunded by the European Union within the framework of the Fourth EU Health Programme. The Department of Pathophysiology and Transplantation, University of Milan, is funded by the Italian Ministry of Education and Research: Dipartimenti di Eccellenza Program 2023 to 2027. The Microscopy Services Laboratory, Department of Pathology and Laboratory Medicine, is supported in part by NIH, National Cancer Institute, Cancer Center Core Support grant P30 CA016086 to the UNC Lineberger Comprehensive Cancer Center. This research has been conducted using the UK Biobank Resource under Application Number 40713 (www.ukbiobank.ac.uk).

Contribution: Y.S., B.C.C., K.N.K., and K.K. performed analyses on the mouse models, blood, and tissues; M.M., F.P., and A.D.S. analyzed data from the HISTORY trial; J.A.B., K.L.W., and N.L.S. analyzed genomic and proteomic data; B.d.L. provided essential reagents for analysis of plasminogen activation; M.J.F. provided critical input on study design and analysis; A.S.W. conceptualized the study, designed experiments, and analyzed data; Y.S. and A.S.W. wrote the manuscript; and all authors reviewed, edited, and approved the final version of the manuscript.

Conflict-of-interest disclosure: B.d.L. is employed by Synapse Research Institute, a not-for-profit member of the STAGO Diagnostic group that holds the patent on calibrated plasmin generation measurements in plasma. A.S.W. has received funding from 501 Ventures to investigate effects of TXA on fibrin structure. The remaining authors declare no competing financial interests.

Correspondence: Alisa S. Wolberg, Department of Pathology and Laboratory Medicine, UNC Blood Research Center, University of North Carolina at Chapel Hill, 8018A Mary Ellen Jones Bldg, 116 Manning Dr, CB 7035, Chapel Hill, NC 27599-7035; email: alisa_wolberg@med.unc.edu.