Extravascular factor IX: a semicentennial narrative review

Abstract

Coagulation factor IX (FIX) is a serine protease that is crucial for normal hemostasis in vertebrates. Unlike other soluble vitamin K-dependent procoagulant enzymes, there is an “extravascular reservoir” of FIX in various tissues. The extravasation of FIX and its subsequent maintenance outside of circulation have long been attributed to its ability to bind collagen IV. This property is conferred by a short, specialized region of its phospholipid-binding gamma-carboxyglutamic acid (Gla) domain. Collagen IV is a nonfibrillar extracellular protein found primarily in basement membranes that assembles into sheet-like supportive structures around blood vessels and epithelia, providing a convenient scaffold for this “hemostatic barrier.” Early seminal investigations from 1980 to 2000 suggested the existence of a large extravascular reservoir containing the majority of a vertebrate organism’s FIX. The subsequent development of a FIX-deficient mouse model, identification of the amino acid residues responsible for FIX’s collagen binding (K5 and V10), and development of enzyme mutants with high and low affinities for collagen IV afforded new insights into the hemostatic importance of this property from 2000 through 2020. Recent interest on the topic has been catalyzed by therapeutic implications for hemophilia treatment. For example, studies have examined the hemostatic and pharmacokinetic effects of enhanced collagen binding on the hyperactive FIX variant (Padua) currently in trials for hemophilia B gene therapy. In this review, we highlight research updates on extravascular FIX since 2020 and present remaining questions to help guide future investigations.

Essentials

• •Coagulation factor IX (FIX) is a key protein that is important for blood clotting.
• •This protein circulates in blood along with other clotting proteins but is also found in tissues outside of blood.
• •Despite decades of research, there are major unanswered questions about extravascular FIX.
• •More research is needed to solve important problems in bleeding, clotting, and hemophilia B.

1. Introduction

Coagulation factor IXa (FIXa) is a vitamin K-dependent serine protease required for normal hemostasis. Insufficiency of enzymatic activity in humans yields the hemorrhagic and arthropathic disease hemophilia B. FIX is synthesized by hepatocytes in the vertebrate liver and undergoes extensive posttranslational modification before secretion into circulation as a mature single-chain protein zymogen composed of 415 amino acids with a molecular mass of approximately 57 kDa. The structural and biochemical characteristics of FIX relevant to human health have been recently and superbly reviewed in additional detail elsewhere [1,2]. Studies of the more common FVIII insufficiency (hemophilia A) throughout the 1940s-1950s revealed that mixing hemophiliac patient plasmas from separate families could correct clotting defects, ultimately leading to the first description of hemophilia B [3].

Most patients with hemophilia B synthesize and circulate dysfunctional FIX. A myriad of different molecular defects lead to these insufficiencies. This translates to heterogenous clinical conditions and severities [4]. Interestingly, anecdotal experience with FIX replacement in hemophiliacs has long demonstrated a complex and unpredictable relationship between the amount of factor replaced and in vivo recovery of FIX clotting activity. Furthermore, the relationship between FIX activity as measured by plasmatic assay (ie, activated partial thromboplastin time) and associated bleeding tendency has also been considered more complicated and unpredictable than other coagulopathies arising from deficiency of a single enzyme.

This increased pharmacokinetic complexity compared with FVIII was the first hint of a clinically significant extravascular reservoir that has since become well established (Figure [5]). As early as the 1960s, pharmacokinetic studies using injection of radiolabeled FIX in healthy and hemophilic human volunteers suggested an extravascular compartment containing about 3 times the amount of circulating FIX observed to stay in plasma [6,7]. With the subsequent development of experimental FIX inhibitors, animal studies reiterated that the relationship between plasma FIX activity and whole-organism bleeding tendency is more unpredictable than for other circulating factors [[8], [9], [10]]. However, it was not until the 1980s that cell culture and animal studies began to describe the mechanism of FIX extravasation. Subsequent mechanistic studies utilizing FIX-deficient mice and engineered enzyme with high or low collagen affinity further elucidated the specific amino acid residues responsible for this property.

Figure.

Schematic diagram of the “extravascular reservoir” of factor IX (FIX). Top: Cross-section of a small blood vessel demonstrates a combination of circulating, plasma-borne FIX as well as collagen-bound FIX associated with the vascular basement membrane. Bottom: The role of the epithelial basement membrane in binding and maintaining an extravascular reservoir of FIX has not yet been defined. Questions are emerging regarding whether and how FIX leaves circulation and encounters this collagenous structure, as well as how it participates in hemostasis. GI, gastrointestinal.

In this review, we aim to first succinctly summarize 40 years of research on the discovery, distribution, and hemostatic significance of extravascular FIX. Next, we provide a more detailed description of work published over the past 5 years. We focus on the basic mechanistic insights provided by several exciting preclinical studies during this period. Finally, we end with a discussion of current knowledge gaps and suggest future directions for research in this field.

2. Seminal Studies (1980-2000)

The first published investigations specifically implicating an extraplasmatic binding site for FIX and FIXa date back to 1983. Essentially contemporaneous reports from research groups at Columbia University (Drs David Stern and John Owen [11]) and the University of Washington (Drs Ronald Heimark and Stephen Schwartz [12]) described binding of FIX zymogen and enzyme to reaction wells containing cultured bovine aortic endothelial cells with apparent augmentation of enzymatic activity. The important and enduring publication of Stern et al. [11] has the noteworthy distinction of demonstrating an exquisite binding specificity for FIX and FIXa. Specifically, they showed no binding competition from the closely structurally and functionally related FX. This experiment was important to demonstrate that the specialized vitamin K-dependent gamma-carboxyglutamic acid (Gla) domain, which is also present on FX, was insufficient on its own to allow for binding. Subsequent publications from the same group suggested preserved FIXa enzymatic activity after binding, as well as increased affinity of the binding site for active enzyme FIXa after addition of exogenous FVIII and FX [13,14]. They also demonstrated a large reservoir of endogenous FIX in primates that was able to be rapidly displaced by infusion of exogenous bovine FIX, suggesting an endothelium- or vessel-bound compartment for FIX [15]. Although Stern et al. [15] reasonably interpreted their findings to suggest that the endothelial cells themselves presented a cell-surface binding site, it would not be for another decade that the extracellular matrix protein collagen IV (presumably secreted in culture by endothelial cells and perivascular fibroblasts) would be identified as the most likely binding partner for FIX [16].

Starting in the 1990s, Dr Darrel Stafford and colleagues at the University of North Carolina (UNC) at Chapel Hill set about identifying the specific binding region on FIX and its extraplasmatic binding partner. Multiple groups were determined to answer the former question, and existing reports had variably implicated the Gla domain and one or both epidermal growth factor (EGF)-like domains [17,18]. In an elegant set of experiments utilizing FVII/FIX chimeric proteins and subsequent point mutations, Dr Stafford’s group [19,20] showed that the Gla domain from FIX uniquely conferred collagen-binding ability, and they subsequently implicated amino acid residues 3-11 with an emphasis on 5, 9, and 10 based on molecular modeling. Their group then used atomic force microscopy to demonstrate binding of FIX to collagen IV directly in vitro and simultaneously provided the first description of the binding domain on the collagen molecule [21]. These descriptions made way for subsequent investigations utilizing mutant Gla domains with higher and lower affinities for collagen IV to assess for the hemostatic importance of this property.

3. Molecular Mechanisms (2000-2020)

The group at UNC Chapel Hill described 2 separate, complementary mouse models of FIX insufficiency that became crucial to understanding the role of extravascular FIX in hemostasis. First, a FIX knockout model lacking FIX protein (ie, cross-reactive material [CRM]-negative) was described [22]. Although this model admittedly does not recapitulate the most common molecular defect in human hemophilia B, it does provide an invaluable tool for isolating the effects of exogenously administered enzyme preparations on hemostasis. Shortly thereafter, a CRM-positive mouse model was developed at UNC by utilizing a dysfunctional human FIX protein [23]. These invaluable models provided the preclinical research substrate for the subsequent 2 decades of work delineating a nuanced relationship between FIX enzymatic activity and collagen IV binding in vivo. Dr Stafford’s group then described the first targeted FIX Gla domain mutants with decreased (K5A) and increased (K5R) collagen affinities. The authors also provided some of the only human immunohistochemical studies ever published using near-normal popliteal artery sections from excess surgical specimens. After this initial description, they expanded their work to include a K5A knockin transgenic mouse model that notably demonstrated impaired hemostasis in vivo but had no effects on clotting in vitro in a chromogenic assay, providing the first direct evidence of an important hemostatic role for collagen IV binding by FIX [24].

Finally, 2 crucial publications in the 2010s examined the efficacy of exogenous Fc-conjugated and unconjugated (wild-type) FIX prophylactic therapies in CRM-negative and positive hemophilia B mice [25,26]. Eftrenonacog alfa is a commercially available recombinant FIX molecule conjugated to the Fc portion of immunoglobulin G, which increases plasma half-life to theoretically decrease the frequency of infusions required for effective prophylaxis in hemophilia B. In a saphenous vein bleeding model in CRM-negative mice, the hemostatic efficacy of eftrenonacog alfa and unconjugated FIX were equivalent at all doses tested, without additional hemostatic potency beyond a point of presumed “saturation” of the extravascular compartment at 150 U/kg. Interestingly, hemostasis was impaired in CRM-positive mice independent of the type of exogenous therapy administered, suggesting that dysfunctional endogenous FIX can compete for collagen IV binding sites. This observation finally presented a unifying explanation for the unpredictable relationship between plasma FIX activity and bleeding tendency observed clinically in persons with hemophilia.

4. Recent Reports (2020-2025)

In 2020, Herrmann et al. [27] directly examined the biodistribution of exogenously administered, albumin-conjugated recombinant FIX (albutrepenonacog alfa) in the tissues and plasma of wild-type rats and FIX-deficient (ie, CRM-negative hemophilia B) mice. Animals were administered large doses of albumin-conjugated FIX (2000 IU/kg) via bolus injection into the tail vein. Blood was serially sampled after intravenous administration. Animals were perfused to remove intravascular FIX, and tissues were collected for imaging (immunohistochemistry), FIX enzymatic activity quantification (chromogenic substrate assay), and protein antigen quantification (ELISA).

The authors detected albutrepenonacog alfa in plasma and all tissues through 72 hours of study. Although plasma levels were highest immediately after intravenous dosing, tissue levels peaked later (1-7 hours). Kidney tissue showed the strongest signal, and there was an accelerated time to peak antigen levels in renal tissues. Relative to plasma, tissue levels continued to increase up to 72 hours. These findings suggest that extravasated FIX remains enzymatically and hemostatically active.

In addition to reinforcing the concept that so-called extended half-life FIX therapies (ie, with protein conjugates) may continue to provide hemostatic benefit after localization to extravascular compartments, the study also demonstrated very rapid extravasation of plasmatic FIX. It was limited by the lack of wild-type FIX administration, the administration of exogenous FIX to CRM-negative animals, and the lack of study of wild-type rodents with species-specific reagents and substrate. Given these constraints, the authors were unable to provide additional insight regarding the dynamics of intra- and extravascular FIX in hemostasis among healthy or CRM-positive hemophilic states.

The same group previously performed whole-body autoradiography experiments on wild-type rats using radioactively labeled formulations of both standard recombinant and extended half-life FIX products [28]. Using macroscopic techniques for assessment of radiography, they demonstrated rapid extravasation of exogenous FIX into tissues, with a predilection for well-vascularized organs (ie, adrenal and lung). These results generally supported and informed their subsequent experiments with direct quantification of antigen and activity in rodent tissues.

In 2023, van der Flier et al. [29] at Sanofi and Invicro used FIX-deficient mice to study the biodistribution of intravenously administered, radiolabeled recombinant FIX and 2 extended half-life conjugates (FIX-Fc and FIX-polyethylene glycol [PEG]). The authors first performed pharmacokinetic assessments on serial plasma samples from treated mice using ELISA and FIX activity assays. They also used a combination of macroscopic (ie, single-photon emission computed tomography/computed tomography imaging) and microscopic techniques to localize and quantify exogenously administered FIX replacement products in FIX-deficient mice. Finally, in a small but highly novel group of immunofluorescence microscopy experiments, they conjugated recombinant FIX or extended half-life FIX molecules to fluorophores for administration to deficient mice. Mice were sacrificed and tissue specimens were counterstained for collagen I or IV with a polyclonal antibody to assess for colocalization in select tissue types.

This report was the first to utilize radiolabeled FIX conjugates to determine the biodistribution of these molecules. By corroborating findings with both micro- and macroscopic detection techniques, the authors showed that exogenously administered FIX accumulated in highly vascularized tissues at interfaces between blood vessels and their associated tissue beds. There was remarkable retention of FIX and FIX-Fc (eftrenonacog alfa) in trabecular bone and joint tissues, leading the authors to conclude that these products might provide better prophylaxis against hemarthrosis in persons with hemophilia.

Two new fundamental insights on extravascular FIX from this paper relate to a) the remarkable difference in biodistribution between PEGylated FIX and unconjugated preparations, and b) the interesting colocalization of type I collagen and FIX in multiple tissues. With respect to the former, the authors suggest that the large hydrodynamic radius of the PEGylated FIX molecule may have prevented efficient extravasation. This steric hindrance and tendency to retain the molecule in the intravascular compartment may explain its long circulating half-life but lack of penetration and retention in target tissues. Since the molecular weight of Fc-conjugated and PEGylated FIX formulas are nearly identical, it also implies a specific charge or other molecular requirement beyond molecular weight alone to efficiently swap FIX between the extra- and intravascular compartments. Wild-type FIX has a molecular weight of approximately 57 kDa, whereas Fc-conjugated FIX (eftrenonacog alfa) and PEG-40-conjugated (nonacog beta pegol) factor replacements each have a molecular weight of approximately 98 kDa. Fc-conjugated and recombinant unconjugated FIX both demonstrated greater and more durable retention in the extravascular space, findings that were most dramatic in large joints. Indirectly, this observation correlates with our own experience using a PEGylated FIX inhibitor that preserved hemostasis but prevented intravascular thrombosis in a porcine model of extracorporeal circulation [30].

With respect to frequent colocalization of FIX and collagen I, the authors provide remarkably little comment on this interesting finding. Using immunofluorescence microscopy, they showed frequent but subtotal colocalization between type I collagen and retained FIX in the sinusoids of the liver and spleen. The pattern was similar to that expected (and observed) for collagen IV in the kidney and bone marrow in the study, in which some but not all type IV collagen-containing blood vessels and other basement membranes costained for both molecules. This subtotal colocalization is the first but not last suggestion of a potential additional different binding partner or requirement for extravascular FIX that has emanated from recent literature.

In the same year, Machado et al. [31] from CSL Behring and Philipps University Marburg elaborated on prior similar studies by Cooley et al. [25,26] by evaluating the hemostatic efficacy of FIX mutants with lower (K5A) and higher (K5R) affinities for collagen IV, both in vitro and in FIX knockout mice. First, they utilized a combination of familiar (conventional clotting and chromogenic substrate assays) and novel (surface plasmon resonance) techniques to evaluate enzymatic and collagen-binding activities, respectively. Finally, they used tail clip and mesenteric arteriole ferric chloride-induced thrombosis models to assess effects on coagulation in vivo using FIX-deficient mice and examined select tissues for FIX and collagen IV colocalization using immunofluorescence microscopy.

The authors reiterated that FIX mutants with differing collagen-binding capacity had no significant differences in activity in vitro. Specifically, they showed similar kinetics of activation by FXIa for all FIX variants. This is an unsurprising but important finding that confirms the importance of collagen binding by FIX to its hemostatic potency, independent of structural or other effects induced by mutation of the responsible amino acid residues in its Gla domain.

The provocative, fundamental, and novel findings from this study arise from the use of surface plasmon resonance, in which no binding to purified collagen IV by FIX could be demonstrated. This finding is complimented again by subtotal colocalization between FIX and collagen IV in tissue specimens using immunofluorescence microscopy. These findings imply that collagen IV binding by FIX may be a more complex process than initially believed. Whether an additional binding partner may exist (ie, collagen I) or another local factor may be required for efficient binding remains to be determined.

In 2024, Leuci et al. [32] provided an expansive report evaluating the binding efficiencies of 4 formulations of FIX to collagens I and IV in vitro and their biodistributions at varying doses in FIX-deficient mice. Notably, their study is unique in evaluating plasma-derived FIX in addition to recombinant and extended half-life conjugates, an important consideration given some surprising findings unique to the plasma-derived product.

Their simple but elegant assessment of FIX binding to adsorbed collagens I and IV deserves discussion. Using purified collagens adhered to a microtiter plate, the authors added FIX preparations to each well and then probed for collagen-captured FIX with a reporter-conjugated anti-FIX antibody. They reported a high and similar degree of collagen IV binding among all FIX preparations, as expected. However, only plasma-derived FIX bound with great avidity to collagen I; the remaining preparations bound less predictably. The authors suggest that there may be a posttranslational modification or other structural explanation for the increased binding of “native” plasma-derived FIX to collagen I, which has previously not been described as a binding partner for FIX. This experiment provided the strongest evidence to date that collagen I plays some role in hemostasis by extravascular FIX.

The authors also provide the most exhaustive immunohistologic assessment of extravasated FIX yet published. Detailed microscopic examinations of liver and knee joint tissues from factor-deficient animals treated with varying doses of the 4 FIX preparations revealed accumulation in both parenchymal and perivascular distributions of the liver, as well as in the synovium of large joints. Remarkably, much lower doses of plasma-derived FIX yielded greater extravascular staining intensity in the liver and synovium compared with extended half-life or recombinant FIX preparations, suggesting that post-translational or other native protein modifications confer enhanced collagen binding.

An additional insight from this report relates to the reliable colocalization of FIX and collagen IV, but not collagen I. Despite their demonstration of binding and retention of FIX by both collagens IV and I in the purified system described, exogenously administered factor did not reliably stain at sites of collagen I in joint and bone tissues. Instead, FIX was found primarily at sites of expected and confirmed collagen IV—specifically, around bone compartments and perforating vessels near large joints.

A 2024 article by Cloesmeijer et al. [33] described a physiologically-based pharmacokinetic (PBPK) model for exogenously administered recombinant FIX and Fc-conjugated FIX with consideration given to the extravascular compartment based on prior data. Using whole-body scale modeling software (PK-Sim), the authors generated distribution models for both a) standard half-life recombinant FIX and b) Fc-fusion protein-conjugated recombinant extended half-life product (eftrenonacog alfa). The model considers multiple extraplasmatic compartments, including myriad tissues, the endothelium, and the interstitial space.

This study is the first published attempt to quantitatively estimate extravascular FIX dynamics in humans using a mechanistic modeling framework rather than direct imaging or tissue sampling. The authors demonstrated that their PBPK model could reliably recapitulate observed plasma pharmacokinetics while simultaneously predicting substantial accumulation of recombinant FIX in extravascular compartments. Notably, the model estimated that the area under the curve for exogenous recombinant FIX bound to collagen IV in the extravascular space was approximately 19-fold higher than that in plasma, supporting the concept of a large and clinically important extravascular reservoir. Tissue-specific simulations further suggested preferential accumulation in organs with higher collagen IV expression, particularly the spleen and lungs, where extravascular concentrations exceeded plasma peaks and declined slowly over time.

Two major insights regarding extravascular FIX emerged from this work. First, the magnitude of the extravascular pool relative to plasma supports prior experimental observations that hemostasis may be maintained even in the absence of measurable circulating FIX activity. This reinforces the concept that tissue-bound FIX may serve as a functional depot available at sites of vascular injury and provides a mechanistic framework for the longstanding observation of discrepancies between plasma FIX activity and bleeding tendency. It also may explain differential protection from hemorrhage across tissue beds, since predicted accumulation of FIX differed by orders of magnitude across tissue types. Second, the authors modeled with and without Fc receptor recycling (ie, Fc-fusion protein-conjugated vs standard recombinant FIX) and showed the feasibility of predicting different conjugates’ behaviors. Indeed, simply acknowledging the long-neglected extravascular depot of FIX in these parameters is a novel and progressive consideration; such high-fidelity PBPK modeling may ultimately decrease reliance on animal studies for preclinical work and support translation of new therapeutics in this space.

In 2025, Westmark et al. [34] from University of Wisconsin-Madison aimed to describe the regions of FIX responsible for extravasation and maintenance in the extravascular space. The authors generated FIX protease mutants affecting distinct exosites known to be responsible for a) heparin binding, b) antithrombin binding, and c) cofactor binding. They also included an active site-inhibited FIX condition utilizing wild-type enzyme pretreated with an irreversible protease inhibitor to evaluate the effects of the active site. In an elegant series of experiments, the authors administered either the active site-inhibited wild-type human FIX or an exosite mutant to CRM-negative hemophilia B and hemophilia A mice. Then, they assessed effects on hemostasis, thrombotic tendency, pharmacokinetics (ie, plasma recovery of FIX), and biodistribution.

The authors’ results unsurprisingly reflect nuanced roles for each exosite in hemostasis, thrombosis, plasma recovery, and/or biodistribution. However, the most striking finding with major implications for the mechanism of endogenous FIX extravasation was increased plasma recovery (and decreased tissue detection) of the FIX mutant with heparin-binding exosite disruption (FIX K126A/K132A). This novel finding implicates the heparin-binding exosite in collagen binding, extravasation, and maintenance of FIX outside of circulation.

The study showed a lack of exogenously administered FIX in brain tissues. This is consistent with prior reports from Herzog et al. [28] and Chang et al. [35]. The authors attribute this finding to maintenance of a blood–brain barrier by the specialized endothelium, pericytes, and microglial cells, which may exclude circulating molecules from the brain’s vascular basement membrane. Additional consideration might be given to the specialized histologic structure of the brain’s parenchyma, which lacks a conventional epithelium and associated basement membrane apart from its thin ependymal ventricular lining.

Interestingly, disruption of the heparin-binding exosite resulted in similar hemostatic activity in vivo using a saphenous vein bleeding model, but in vitro enzymatic activity was decreased substantially in prior work published in 2015 from the same group [36]. The latter observation was attributed to impaired cofactor affinity, providing an additional layer of complexity when interpreting the hemostatic impact of changes to the domains responsible for FIX extravasation.

Finally, in an exciting culmination of this recent work, a 2025 report from Aaen et al. [37] describes synthesis and rodent testing of a hyperactive FIX mutant (Padua, R338L) with either decreased or increased collagen binding via K5 point mutations (K5A or K5R, respectively). FIX Padua is a well-characterized gain-of-function mutant that causes thrombophilia in humans [38]. This hyperactive FIX molecule has become the dominant candidate gene product for ongoing trials in hemophilia B gene therapy due to its increased activity, allowing for improved hemostatic function in vivo with lower concentrations of circulating protein [39]. Prior to this effort, the authors had described a FIX Padua molecule conjugated to recombinant human serum albumin; this mutant albumin was optimized for neonatal Fc receptor interaction, which extended plasma half-life significantly [40,41].

The authors ultimately performed pharmacokinetic studies of wild-type and engineered human albumin-FIX Padua conjugates in mice with normal circulating FIX levels and then evaluated the effect of collagen-binding capacity by comparing K5A and K5R mutant pharmacokinetic profiles. As expected, favorable interactions with the endogenous neonatal Fc receptor increased the circulating half-life of exogenously administered FIX. They followed with a similar set of experiments using FIX-deficient, CRM-negative hemophilia B mice. Unsurprisingly, FIX conjugates with decreased collagen IV binding (K5A) resulted in greater early circulating FIX activity in CRM-negative mice compared with K5R conjugates with greater collagen affinity. To confirm the hypothesis that this finding arose from increased extravasation of exogenous FIX via collagen binding, the authors used a combination of ELISA and immunofluorescence microscopy to quantify FIX in tissue homogenates from treated mice. They demonstrated increased maintenance of K5R-Padua in the liver, kidneys, lungs, and joints 4 days after injection.

Finally, the authors used a tail-clip hemostatic challenge among treated mice on day 4 after injection of engineered conjugates. This blood loss assay confirmed that K5R promoted improved hemostasis commensurate with its increased residence time in vivo. By comparison, K5A had an early, higher peak of plasma activity indicative of poor extravasation into and binding of collagen in tissues. This led to ineffective repletion of the extravascular hemostatic depot and impaired hemostatic efficacy during tail-clip challenge.

Interestingly, this report was the first to note that K5 mutants bound phospholipid with different avidities. The authors created an ELISA-based assay in which phospholipid vesicles bound to wells were used to detect binding of their engineered variants at differing concentrations. Although this finding is not entirely surprising given the known phospholipid-binding role of the associated Gla domain, the novel experiments were the first to confirm that apparent in vitro activity differences of the variant enzymes were likely due to their differential affinities for phospholipids added during clotting assays. It also underscored the bifunctional importance of the K5 residue in both collagen and phospholipid binding.

5. Remaining Questions

These studies have expanded our insight into the nature and hemostatic importance of FIX’s collagen binding both in vivo and ex vivo using rodent models and plasma-based testing (Table [27,29,31,33,34,37]). In particular, the last 5 years have seen improvements in our understanding of the pharmacokinetics and dynamics of exogenously administered FIX and its interactions with the extravascular compartment. However, these studies now raise a number of fundamental questions that are crucial to fully describing and exploiting FIX extravasation for human health and to holistically enrich our understanding of hemostasis and biology [42].

Table.

Brief summary of recent publications on extravascular factor IX.

Journal, year [reference]	Authors	Primary techniques and assays	Major findings
Journal of Thrombosis and Haemostasis, 2020 [27]	Herrmann S, Doerr B, May F, Kuehnemuth B, Cherpokova D, Herzog E, et al.	Autoradiography, ELISA, and colorimetric activity assays on tissue homogenates from rodents.	Exogenously administered FIX products accumulate in rodent tissues and maintain enzymatic activity.
Blood Coagulation and Fibrinolysis, 2023 [29]	van der Flier A, Hong V, Liu Z, Piepenhagen P, Ulinski G, Dumont JA, et al.	SPECT/CT, immunohistochemistry, and plasma recovery in mice.	Exogenously administered FIX accumulates in joint and bone space; FIX often colocalizes with collagen I.
Thrombosis and Haemostasis, 2023 [31]	Machado SK, Peil H, Kraushaar T, Claar P, Mischnik M, Lind H, et al.	Decreased (K5A) and increased (K5R) collagen-binding mutants, surface plasmon resonance, mouse models of thrombosis (ferric chloride) and hemostasis (tail clip).	No detection of collagen IV binding by FIX using surface plasmon resonance. Increased collagen-binding results in tissue deposition and promotes hemostasis but does not affect FIXa activation of FIX in vitro.
CPT: Pharmacometrics & Systems Pharmacology, 2024 [33]	Cloesmeijer ME, Sjogren E, Koopman SF, Lenting PJ, Cnossen MH, Mathot RAA, et al.	Physiologically-based pharmacokinetic modeling.	Model predicts differential accumulation of exogenous FIX in tissues; orders of magnitude in difference among tissue concentrations based on collagen content.
Journal of Thrombosis and Haemostasis, 2025 [34]	Westmark PR, Annis DS, Torres B, Misenheimer TM, Schwartz BS, Hutson PR, et al.	Human FIX variants with targeted mutations involving exosites (or an active site inhibitor). Saphenous vein bleeding and ferric chloride clotting assays in mice.	FIX’s heparin-binding exosite is involved in extravasation and collagen binding.
Nature Communications, 2025 [37]	Aaen KH, Testa MF, Nilsen J, Tarantino R, Canepari C, Benedusi M, et al.	Hyperactive human FIX variant (Padua) with collagen-binding affinity mutations (K5A, K5R) and engineered albumin conjugates. Experimental molecules administered to FIX-competent and -deficient mice challenged with tail-clip and FIX recovered from plasma and organ homogenates (ELISA).	Increased collagen affinity confers longer residence time in circulation and tissues and improved hemostasis. FIX collagen-binding domain modifications also predictably affect phospholipid binding and apparent activity in vitro.

CT, computed tomography; FIX, factor X; SPECT, single-photon emission computed tomography.

First, several authors have raised the possibility of additional extravascular binding partners for FIX beyond collagen IV. Studies have demonstrated inconsistent or incomplete FIX-collagen IV colocalization in animals and humans [25,29,31]. This aligns with our own observations in perfused rats, in which there was FIX-collagen IV colocalization around some vascular structures but not others [43]. One possibility is that there is an additional extracellular matrix protein binding partner, for example, collagen I. The most compelling data to suggest that collagen I binds FIX are described in 2023 by Leuci et al. [32], in which their immobilized collagen-binding assay demonstrated strong binding of wild-type, plasma-derived FIX to collagen I. It is plausible that FIX binds both collagens: although the supramolecular organization of collagens I and IV are distinct (fibrillar vs laminar), both collagens are evolutionarily ancient and present in all animals [44]. Most importantly, collagen IV’s likely binding domain for FIX is within its triple helical collagenous domain; there are analogous domains in collagen I and other collagens [21,45].

Another possible explanation for this frequent but inconsistent colocalization is that collagen IV is required but not sufficient for FIX binding; that is, a second factor is required for efficient binding. In one experiment attesting to the nuances required for efficient binding, Machado et al. [31] were unable to demonstrate direct recombinant FIX-collagen IV binding in vitro using surface plasmon resonance. Given the structural complexity of the FIX molecule and its known additional exosite binding partners, it would not be surprising that a cofactor (ie, endogenous heparinoid) or specific local milieu (ie, sodium or other ion concentration) might be required to confer efficient Gla domain binding.

A second area of opportunity is improving and standardizing techniques for the measurement of extravascular FIX. Existing reports have utilized ELISA or chromogenic substrate assays to evaluate FIX antigen and activity levels in tissue homogenates, respectively. However, ELISA is unable to measure enzymatic activity, and chromogenic or fluorometric substrate-based enzyme activity assays are often species-specific and are limited by lack of a commercially available FIX-specific substrate. Semiquantitative techniques for whole-tissue comparison, such as single-photon emission computed tomography/computed tomography and autoradiography, provide important macroscopic insights but necessarily rely on provision of exogenous labeled FIX. All techniques must compensate for a background “blood pool” of circulating FIX in tissues unless they are acquired after vascular perfusion (as may be acquired premortem in animal models, or theoretically in carefully curated human specimens that have undergone ex vivo perfusion). Therefore, much of our current understanding of extravascular FIX’s biodistribution is based on FIX-deficient animal models. Future studies should delineate optimal assays for detection and quantification of tissue-specific FIX antigen and activity levels, with careful consideration given to “contamination” by FIX in circulating plasma and species-specific differences.

Because of the limitations of vascular perfusion and need for surgical or pathologic study of tissue homogenates for direct quantification of extravascular FIX, data from human tissues are extremely limited. Thus far, human specimens have only been analyzed sporadically for nonquantitative immunohistochemical studies [25,46]. The amino acid sequence of FIX is similar across mammals, with over 80% homology of mouse and human sequences and preservation of the crucial K5 and V10 residues that confer collagen binding in mammalian model species [47]. Furthermore, the most commonly used mouse model is CRM-negative, which does not recapitulate the most common lesions observed in persons with hemophilia. Therefore, in future studies, specimens from human donors should be considered for analysis wherever possible.

Although collagen IV is an important constituent of the vascular endothelium in vertebrates, there is also an abundance of collagen IV in other nonvascular basement membranes. All studies of extravascular FIX have thus far focused on the vascular endothelial basement membrane. Therefore, there is currently a major knowledge gap in how nonvascular basement membranes bind FIX. Questions abound regarding how FIX might bind to collagen remote from circulation, as well as its equilibrium with circulation and importance to hemostasis. Microscopy studies may help elucidate the size and location of these additional depots of FIX remote from the vasculature.

Finally, the clinical implications of this extravascular reservoir are still emerging. Our evolving and incomplete understanding of the role of extravascular FIX in maintaining hemostasis makes it intriguing but impossible to make declarations about its importance in healthy persons and those with hemophilia. However, the size and hemostatic potency of this reservoir (as well as its demonstrably differential response to factor replacement therapies) in animal studies and in vitro suggest an important role in preventing and stopping clinically relevant bleeding. Future investigations in both healthy individuals and those with hemophilia and in animal models should consider assessing hemorrhagic tendency with real-world hemostatic challenges, instead of relying on plasmatic factor activity to estimate bleeding liability.

Nearly 50 years of careful study of extravascular FIX has yielded incredible insight into this novel property and its hemostatic significance with major implications for hemophilia, normal hemostasis, coagulation biology, and beyond. Further investigations may yield additional breakthroughs with associated implications for human health.