Introduction
Tissue factor (TF) is the primary factor responsible for initiating the coagulation
cascade by forming a complex with factor (F) VII and FVIIa, which can activate FX
and FIX, subsequently generating thrombin. Besides, TF is also involved in other cellular
processes, including cell survival,[1] angiogenesis,[2] metastasis formation, and tumor growth.
Different forms of TF circulate in blood either as a component of blood cells or soluble
plasma protein or associated with extracellular vesicles (EVs) which are membrane-enclosed
submicronic vesicles released by all types of cells that carry molecular cargo from
their parental cells. Elevated levels of these circulating forms have been detected
in various pathological conditions, such as cancer,[3]
[4] infectious disorders,[5]
[6] heparin-induced thrombocytopenia,[7] sickle cell disease,[8]
[9]
[10] acute myocardial infarction,[11] and preeclampsia.[12] Interestingly, a link has been found between increased EVTF activity in a large
spectrum of pathological contexts and the disease severity, such as the stage of cancer,[4]
[13] or patient mortality,[14]
[15]
[16]
[17]
[18]
[19] as recently reviewed.[20] In addition, increased EVTF activity was also correlated with the occurrence of
disseminated intravascular coagulation (DIC)[21]
[22]
[23] or thrombotic events, including venous thrombosis, in patients suffering from cancer[24]
[25]
[26]
[27] or COVID-19.[5]
[28]
In the present review, after a brief overview of the different forms of TF detected
in circulating blood and their mechanisms of biogenesis, we address the available
methodologies, including in-house and commercial assays that have been published to
measure TF in clinical samples, highlighting their main preanalytical and analytical
characteristics, strengths, and limitations. Finally, we will discuss potential next
steps toward the application of TF measurement in clinical laboratories.
Different Forms of Circulating TF
TF is a transmembrane protein composed of three domains: an extracellular domain of
219 amino acids that can bind to factor VII, a transmembrane domain of 23 amino acids,
and a cytosolic domain of 21 amino acids that is involved in signal transduction.[29] Different forms of TF circulate in the blood, either as a component of blood cells
and EVs or as a soluble plasma protein. Under physiological conditions, most of the
TF associated with the cell or EV membrane is detectable in an encrypted nonactive
form. This encryption is related to posttranslational suppression of TF procoagulant
activity by the absence of a disulfide bond in the C-terminal region between unpaired
cysteine residues 186 and 209, which changes the conformation of the TF to a nonfunctional
form.[30] The TF procoagulant activity became detectable after decryption, which implied both
the formation of a disulfide bond between unpaired cysteine residues by protein disulfide
isomerase[31] and the exposure of anionic phospholipids on the membrane surface,[32] which involved an increase in the intracytoplasmic concentration of calcium in cells.
This negatively charged phospholipid surface is required for maximum TF activity.[33]
TF is largely expressed in several organs, including the brain, lung, kidney, and
placenta.[34] To ensure a hemostatic barrier in the case of vessel injury,[35] TF is constitutively expressed by cells covering vessels such as fibroblasts of
the adventitia and vascular smooth muscle in the subendothelium. Moreover, monocytes
and endothelial cells can express TF after stimulation by lipopolysaccharide[36] or by proinflammatory cytokines.[37] The presence of TF on neutrophils[38]
[39]
[40]
[41] and platelets[42]
[43]
[44] is controversial. Several studies have suggested that it could result from contamination
or molecular exchange with monocytes.[45]
The initial concept of TF serving predominantly as a hemostatic envelope encapsulating
the vascular bed has been challenged by the observation that the human plasma may
form TF-induced thrombi, corroborating the notion of “blood-borne TF.”[46] Indeed, in addition to the tissue- and cell-associated compartments, TF is also
mainly detectable in two forms, one associated with EV and the other as a soluble
form generated by alternative splicing (asTF) containing only the extracellular domain.
This soluble form of TF is able to slowly bind to factor VIIa, but the amidolytic
properties of the asTF–FVIIa complex are substantially reduced because of the absence
of a plasma membrane with anionic phospholipids to facilitate the spatial accumulation
of coagulation factors.[47] Indeed, endothelial cells can release asTF after stimulation with tumor necrosis
factor-α and interleukin-6. This soluble form of TF becomes a procoagulant only in
the presence of phospholipids.[48] Besides, EVs could be another source of “blood-borne TF.” Several EV subsets, such
as monocytes,[49] endothelial cells,[37]
[50] and tumor-derived EVs,[51] have been reported to carry functional TF.[52] In addition to the origin of cells triggering their formation, the composition of
EVs can be modulated by stimuli that induce their biogenesis. Accordingly, the expression
of TF on EVs increases due to factors such as proinflammatory cytokines[37] and individual habits such as smoking.[53] Conversely, the EVTF decreases by pharmacological intervention such as inhibitors
of the protein convertase subtilisin/kexin type 9[54] and the direct oral anticoagulant apixaban.[55]
However, since the first demonstration of the blood-borne TF,[46] several controversies have been highlighted in the literature. First, an increase
in EVTF activity was not consistently observed within the same pathophysiological
context.[15]
[17] Second, the presence[56] or absence[57]
[58] of EVTF in healthy individuals has been an active matter of debate. However, it
is increasingly accepted in the scientific community that circulating TF is present
in low quantities in healthy individuals. Most of these discrepancies are probably
due to methodological heterogeneity. Indeed, the preanalytical and analytical conditions
largely differ among the numerous methods used to measure circulating TF.
Methods to Measure TF
Based on the different forms detectable in clinical samples, the following sections
will review the main in-house and commercially available assays to measure TF detectable
in whole blood (red section), in platelet-poor plasma (yellow section), or in EVs
(blue section; [Table 1]). This article will discuss the main preanalytical and analytical parameters that
impact the results and assay interpretation, highlighting their strengths and limitations.
Within each section, methods have been divided into two categories: TF antigenic assays
(A) and TF functional assays (F). For each assay, the principle of the assay, the
preanalytical step, an estimation of the duration, and the commercialized or in-house
version of the test are listed in [Table 1]. Given the multitude of assays that have been developed, only the most representative
references were selected for each assay.
Table 1
Main methods used to measure TF
|
Method
|
Reference
|
Preanalytical step
|
Sample preparation duration (min)
|
Reagent incubation duration (min)
|
Sensitivity
|
Specificity
|
In-house or commercialized assay
|
|
Whole blood
|
|
F
|
Thromboelastography
|
[59]
[60]
[61]
|
NA
|
NA
|
<30
|
Unknown
|
No aTF
|
ROTEM,[a]
[b] TEG,[a]
[b] Sonoclot[a]
[b]
|
|
Clotting assay
|
[10]
|
NA
|
NA
|
<30
|
Unknown
|
aTF
|
In-house
|
|
TiFaCT
|
[62]
|
NA
|
NA
|
20
|
70 fM
|
aTF
|
In-house
|
|
Platelet-poor plasma
|
|
F
|
FXa generation
|
[64]
|
Centrifugation
|
30
|
45
|
2,000 fM
|
No aTF
|
ACTICHROME TF[c]
|
|
FIIa generation
|
[65]
|
Centrifugation
|
20
|
<30
|
Unknown
|
No aTF
|
CT[c]
|
|
A
|
ELISA
|
[67]
|
Centrifugation
|
10
|
≈180
|
Imubind: 1,400 fM§
Quantikine: 20 fM
Hum SimpleStep: 103 fM
|
NA
|
IMUBIND TF,[c] Quantikine,[c] Human SimpleStep[c]
|
|
[4]
[91]
|
Centrifugation
|
15
|
≈180
|
1,400 fM
|
NA
|
In-house
|
|
Extracellular vesicles
|
|
F
|
FXa generation
|
[72]
|
Centrifugation
|
120
|
160
|
3 ± 1 fM
|
aTF
|
CY-QUANT MV-TF Activity[c]
|
|
[4]
[92]
|
Centrifugation
|
90
|
105
|
12 ± 1 fM
|
aTF
|
In-house
|
|
[93]
|
Centrifugation
|
60
|
105
|
Unknown
|
No aTF
|
In-house
|
|
[94]
|
Immunocapture
|
Overnight
|
240
|
28 fM
|
aTF
|
ZYMUPHEN MP-TF ACTIVITY[c]
|
|
[69]
|
Immunocapture
|
60
|
150
|
Unknown
|
aTF
|
In-house
|
|
FIIa generation
|
[71]
|
Centrifugation
|
30
|
20
|
15 fM
|
HFVIIai
|
In-house
|
|
[70]
|
Centrifugation
|
60
|
30
|
Unknown
|
aTF
|
In-house
|
|
[95]
|
Centrifugation
|
30
|
NA
|
Unknown
|
No aTF
|
In-house
|
|
[55]
|
Centrifugation
|
30
|
NA
|
Unknown
|
No aTF
|
In-house
|
|
Fibrin generation
|
[96]
|
Centrifugation
|
30
|
30
|
Unknown
|
aTF
|
In-house
|
|
A
|
Single EV flow cytometry
|
[81]
[97]
|
Centrifugation
|
60
|
≈30–150
|
*
|
*
|
In-house
|
|
[98]
|
NA
|
NA
|
≈30
|
*
|
*
|
In-house
|
|
Beads-based flow cytometry
|
[76]
|
Immunocapture or centrifugation
|
NA
|
Overnight
|
*
|
*
|
MACSPlex EV[c]
|
Abbreviations: aTF, tissue factor–blocking antibody; HFVIIai: Human Factor Vlla Inactivated;
NA, not applicable.
Notes: The column entitled “sensitivity” lists the limit of detection of each assay
when this characteristic is known. This needs to be interpreted with caution in the
absence of a universal TF or EV-TF standard.
When endpoint of the calibration curve was used to estimate the sensitivity of the
assay, the symbol “§” was added.
When the multicenter study demonstrates a lower specificity and sensitivity of antigenic
assays compared with functional assays, the symbol “*” was added.
a Food and Drug Administration (FDA) approval.
b European Conformity (CE) approval.
c Research use only.
Red: Whole blood based assays.
Yellow: Platelet-poor plasma based assays.
Blue: Extracellular vesicles based assays.
Whole Blood
More than 20 years ago, different functional assays have been developed to measure
TF activity in whole blood. Among them, thromboelastography[59]
[60]
[61] measures clot formation around a thin wire probe used as a sensor. The system is
commercialized by various manufacturers: for example, ROTEM (TEM International GmbH,
Munich, Germany), TEG (Haemonetics, Boston, Massachusetts, United States), and Sonoclot
(Sienco, Inc., Morrison, Colorado, United States). All these assays are Food Drug
Administration (FDA) approved to assess hypercoagulable tendencies but not yet CE-IVDR
(European In Vitro Diagnostic Medical Devices Regulation) labeled. Among several parameters
deduced from the thrombus formation curve, the lag time has been shown to be correlated
with the amount of TF present in the samples.[60] However, the absence of a TF-blocking antibody limits the specificity of these assays.
Indeed, procoagulant activity can be either TF dependent or independent. Consequently,
the use of a TF-blocking antibody enables the determination of the activity specifically
dependent on TF, usually after subtraction of the activity of the control condition
obtained by performing the same assay with an isotype antibody. This is incorporated
in the tissue factor clotting time (TiFaCT) assay which measures the time of whole
blood clot formation after recalcification in the presence or absence of a TF-blocking
antibody.[62] Similarly, another clot formation assay including a TF-blocking antibody recommends
carrying out a freeze–thawing cycle during the preanalytical step, to lyse all cellular
components and decrypt any TF in the cell membrane.[10] In summary, the main advantage of these whole blood methods is the almost complete
absence of a preanalytical step, while it implies that the assays must be done quickly
to avoid any artifactual activation of the cells without the possibility of storage.
Moreover, the sensitivity of these assays can be limited because of the presence of
TF inhibitors such as tissue factor pathway inhibitors (TFPIs). Indeed, TFPI is known
to be one of the main physiological inhibitors of TF by forming a complex with TF,
FVIIa, and FXa. The choice to have an assay not sensitive to TFPI and/or to increase
the sensitivity by blocking TFPI is a matter of discussion. On one side, the use of
an anti-TFPI antibody was shown to increase the TF activity[63] and therefore the sensitivity of the assay. On the other side when blocking TFPI,
the assay will no more integrate the impact of the main TF physiological inhibitor.
Platelet-Poor Plasma
TF can also be measured in platelet-poor plasma (PPP). Several commercial and in-house
assays including both antigenic and functional methods have been developed ([Table 1], yellow section). Regarding functional methods, ACTICHROME[64] (BioMedica Diagnostics, Windsor, Canada) is a commercial plasma TF activity assay
in which FVIIa, FX, and a chromogenic substrate for FXa are added to the plasma. Importantly,
no TF-blocking antibody was used to confirm the specificity of the assay. In addition,
an automated thrombin generation assay has been developed called calibrated automated
thrombinogram[65] which can be performed on a commercialized instrument like ST Genesia (Stago, Asnières-sur-Seine,
France). This method assesses the dynamics of thrombin generation. According to a
standard curve, thrombin generation is detected by the cleavage of a fluorogenic substrate
in the sample, and the lag time can be used to monitor the levels of circulating TF
in plasma.[60]
These assays can be performed quite rapidly (< 1 hour). However, several concerns
were raised regarding a limited sensitivity probably due to the presence of TF inhibitors
such as TFPI in the plasma and a limited specificity especially when these assays
are performed without TF-blocking antibodies and a high concentration of FVIIa.
Regarding antigenic tests, several commercial assays including Imubind (BioMedica
Diagnostics), Quantikine (R&D Systems, Minneapolis, Minnesota, United States), and
Human SimpleStep (Abcam, Cambridge, United Kingdom) or in-house enzyme-linked immunosorbent
assays (ELISAs) have been developed to quantify the TF antigen in the PPP ([Table 1]). They most commonly used an anti-TF monoclonal antibody as a capture antibody and
either a monoclonal or polyclonal anti-TF antibody as a detection antibody, which
should increase the specificity of the method. The choice of the antibody clone is
a key point that directly impacts the specificity of the ELISAs. Indeed, one previous
study attributes false-positive TF detection to the cross-reactivity of the antibody
with non-TF proteins.[66] Moreover, another issue is that most ELISAs detect both active and nonactive forms
of TF such as the asTF, which has little or no procoagulant activity. Consequently,
a large variation of the plasma TF concentration has been reported in healthy subjects
between in-house methods[66] and commercial ELISA,[67] with variations ranging from picomolar to nanomolar levels.
Overall, despite the ease of the preanalytical step to obtain PPP, plasma-based assays
present limited analytical performances.
Moreover, it is important to note that all the previously commercialized assays are
for research use only and do not yet have FDA approval or CE-IVDR labeling.
Extracellular Vesicles
Given that active circulating TF in plasma is primarily found on EVs,[68] various methods (functional and antigenic) have been developed to measure EVTF.
Regarding the functional assays, FXa, FIIa, and fibrin generation assays have been
developed, as illustrated in [Fig. 1]. Briefly, EVs contained in a sample are purified by either high-speed centrifugation
or immunocapture. Then, their specific procoagulant activity is measured by incubation
with a reaction mixture that contains coagulation factors and calcium, with or without
a TF-blocking antibody. Next, the reaction is blocked by a calcium chelator such as
EDTA, and a specific fluorogenic or colorimetric substrate of FXa or FIIa is used
to reveal the reaction. For fibrin generation assays, turbidimetry allows monitoring
the clot formation. A large spectrum of assays has been published,[4]
[69]
[70]
[71]
[72] and the most representative are illustrated in [Table 1]. Among them, several differences in the protocol were observed regarding the preanalytical
or analytical steps.
Fig. 1 Global method for measuring the TF-dependent procoagulant activity of EVs. a, activated;
EDTA, ethylenediaminetetraacetic acid; EVs: extracellular vesicles; f, factor; TF,
tissue factor.
EV purification has been identified as one of the major sources of variability, and
ultracentrifugation is the most likely cause of intra-assay variability. Previous
studies have shown that ultracentrifugation can be influenced by the rotor type, centrifugation
speed (g force), temperature, brake use, and centrifugation duration.[72] For example, one study indicated that increasing the centrifugation force from 20,000 g to 100,000 g does not result in an increase in EVTF activity in plasma issued from whole blood
stimulated with LPS.[73] Conversely, another study recommended applying a centrifugation force of 100,000 g
[74] after demonstrating that both large and small EVs exhibit TF-dependent procoagulant
activity in a proportion that varies according to the pathophysiological context.
Additionally, ultracentrifugation steps are time-consuming and not available in most
laboratories. Therefore, providing access to an EV preparation strategy independent
of ultracentrifugation is a major challenge for measuring EVTF activity in routine
biological laboratories. Already, more than 20 years ago, a hybrid assay was developed
combining the capture of EVs on a plate by a TF antibody that does not interfere with
its activity and the measurement of FXa generation. This test was commercialized as
the Zymuphen MP-TF (Hyphen Biomed, Neuville-sur-Oise, France). The overnight incubation
needed to capture EVs is a limitation of using this assay in clinical practice. More
recently, an immunocapture strategy based on immunomagnetic separation using beads
coated with CD29 and CD59 antibodies was proposed to enhance the repeatability of
the assay.[69]
Other differences among EVTF functional assays rely on the concentration of the coagulation
factors and the choice of the TF-blocking antibody which can be a source of variability.
Indeed, in one study, the SBTF1 antibody displayed a better performance than the HTF1
antibody.[72] The SBTF1 antibody has been incorporated in the recent commercial assay as CY-QUANT
MV-TF Activity (Stago). The assays also differ in their duration which is largely
influenced by the EV preparation steps and the incubation time ranging from less than
1 hour to more than 4 hours. One recent study proposed shortening the assay duration
by performing a single centrifugation of the plasma to obtain the EV pellet, without
the need for additional washing steps.[71]
Regarding the antigenic assays, the most commonly employed method to detect TF on
EVs is flow cytometry (FCM) by either single-EV FCM[75] or bead-based FCM.[76] Besides TF measurement, FCM remains the most commonly used method to enumerate EV
subsets in patients. This technology is available in many research and clinical laboratories
and theoretically has a high potential for EV analysis. FCM detection relies on the
combination of a scatter light profile and the expression of specific antigens detected
by fluorescence. The main advantage of this method is that single-particle analysis
allows the detection of different EV subsets on a quantitative basis. Moreover, unlike
functional tests, the FCM technique allows analysis without purification of the EVs,
which saves a significant amount of time. Despite these strengths, the current flow
cytometer lacks performances to reliably measure EVTF. This can be explained because
a large proportion of EVs has a small size below the detection limit of the instrument
and a low antigen density with only a few molecules of TF present on each EV. Alternatively,
bead-based FCM assays were developed to capture EVs with tetraspanin antibodies (CD9,
CD63, and CD81), which allow to catch small EVs and increase the fluorescence signal.
This method is commercialized as MACSPlex EV Kit IO (Miltenyi Biotec, Bergisch Gladbach,
Germany). However, an overnight incubation is needed to capture EVs which compromises
the use of this technique in clinical laboratories. Another recent development to
improve the performance of FCM to detect EVTF in plasma samples highlights the importance
of the fluorochrome-to-protein ratio as a key parameter in addition to the antibody
clone.[77]
As for the plasma assays, the commercialized EV-based assays are for research use
only so far and do not yet have FDA approval or CE-IVDR labeling.
Besides FCM, other antigenic assays, such as Western blot,[76] superresolution,[78] or confocal[79] microscopy, are available, but they are labor intensive and not amenable to large-scale
use in clinical studies.
Assay Comparison
First of all, to our knowledge, there are no studies that compare whole blood, plasma,
and EVs methods to detect circulating TF simultaneously. Besides, there are only a
few monocentric studies that compare the analytical performances of TF detection from
PPP or EVs. One of them evaluates the performances of four commercial ELISAs in PPP
compared with an FXa generation assay on EVs.[80] They conclude that ELISA methods are not able to detect TF in plasma obtained from
LPS-stimulated whole blood compared with the functional assay performed on EVs, showing
a poor sensitivity of the ELISAs methods. The other studies also show that functional
assays performed on EVs have the best analytical performance in terms of specificity
and sensitivity compared with antigenic assays such as commercial ELISA[23]
[73]
[80]
[81] and bead-based FCM.[82]
Additionally, several monocentric studies compare the performances of functional assays.
Two studies compared an in-house FXa and a thrombin-generation assay to a commercial
TF immunocapture assay and reported a lower specificity and sensitivity of the commercial
assay.[83]
[84] Recently, a study compared the thrombin-generation assay published by Østerud et
al[71] to the FXa-generation assay developed by the Mackman team. These two assays allow
the measurement of EV-TF in a specific and reproducible manner.[85]
Recently, a multicentric study allowed for the first time to compare the analytical
performances of 27 EVTF assays (18 functional and 9 FCM assays).[86] The aim of this study was to compare the specificity, the sensitivity, and the repeatability
of the assays. The specificity was assessed by using EV-derived HAP-1 cell lines,
wild type or knockout for TF. The sensitivity was evaluated using three different
models of EVs: (1) two levels of EVs derived from wild-type HAP-1 cell, (2) PPP obtained
from whole blood with or without LPS stimulation,[37] and (3) two levels of EVs derived from human milk.[87] The repeatability was estimated by the coefficient of variation since each type
of sample was measured in triplicate. The results of this study demonstrate that the
evaluated FCM methods have less specificity than most functional assays because they
were not able to distinguish samples that contained EVs with or without TF expression.
Regarding the comparison of the functional assays, the use of a TF-blocking antibody
or specific immunocapture improved the specificity and sensitivity of the assays,
regardless of the principle (FXa, FIIa, or fibrin generation assays), in agreement
with the recommendations previously established by expert teams.[88] Moreover, it appears that the preanalytical method used to purify EVs affects the
intra-assay reproducibility. Indeed, the assays that use immunocapture instead of
classical centrifugation have a lower coefficient of variation. Finally, this recent
study demonstrated that despite the use of a common recombinant TF as a calibrant,
which should have prevented a significant impact of variables affecting the analytical
step, considerable variation in TF activity measurement remains between the different
functional assays. This encourages the continuation of the efforts to improve the
standardization of these assays with the aim of transposing the measurement of EVTF
in clinical practice.
Conclusion and Perspectives
As addressed in the previous section, each assay has advantages and limitations. In
particular, serious concerns have been raised in terms of sensitivity, specificity,
and reproducibility regarding the measurement of TF activity in whole blood and plasma
to ensure robust and reproducible data between laboratories. These limitations may
explain why EVTF measurement methods have been significantly developed over the past
decade. Additionally, substantial evidence has accumulated in recent years demonstrating
the correlation between EVTF levels and patient outcomes in various diseases.[20] However, there is currently no ideal assay.
Future directions for progress are based on different strategies that are not exclusive:
-
Better define the impact of preanalytical variables on data reproducibility and whether
the measurement of TF is hampered by freeze–thawing deserves more attention. As a
correlate, this variability can be overcome by developing a preanalytical step independent
of centrifugation.
-
Standardize reagents and harmonize protocols that emerge from multicentric assays
with the best analytical performance in terms of specificity and sensitivity.
-
Move toward an automated version of these assays.
-
Improve the FCM performance, in terms of sensitivity and calibration strategies, to
adapt FCM analysis to reliable EVTF measurements.
-
Develop an international standard of TF adapted to EV, either for functional or antigenic
assays.
-
Compare simultaneously several assays not only for their analytical performances but
also their clinical performances.
-
Combine circulating TF with other biomarkers of coagulation, such as D-dimer, in scoring
systems, as proposed, for example, to diagnose DIC in patients with solid cancer[23] and acute leukemia.[89]
The development of these assays aims ultimately to provide new tools for patient management.
For this, several characteristics related to the assays must be evaluated, including
analytical and clinical performances. Several studies have evaluated the analytical
performance of the tests,[69]
[71]
[72] and a recent multicenter study allowed for the first time to compare them.[86] Additionally, the clinical performance of some of these tests has also been evaluated
in various conditions[20] such as COVID-19[5]
[90] which show an association between the EVTF activity level and the occurrence of
thrombosis. However, for hospital use, it is necessary to proceed with the evaluation
of the performance of an industrial test. These data are required to compile a dossier
for FDA approval and/or a CE-IVD label for a specific clinical application.
In conclusion, the measurement of TF-dependent procoagulant activity, especially from
EVs, is a promising biomarker for predicting the occurrence of thrombosis and stratifying
disease severity in different contexts, such as cancer and infectious diseases, to
personalize anticoagulation treatment and improve patient care. However, efforts must
be made to standardize and automate functional assays to measure EVTF in clinical
practice.
