Biomarkers for the diagnosis of deep vein thrombosis

Abstract

Venous thromboembolic disease (VTE) remains a significant source of morbidity and mortality. As non-specific subjective complaints and a paucity of objective clinical examination findings complicate the diagnosis of both deep venous thrombosis (DVT) and pulmonary embolism, diagnostic modalities remain essential. Compression ultrasound remains the gold standard for DVT diagnosis. Reliable imaging is not always available making a serologic diagnosis, or biomarker, highly desirable. While D-dimer, a highly sensitive biomarker, is useful for excluding acute VTE, it lacks the specificity necessary for diagnostic confirmation. As such, ongoing research efforts target and support the utility of alternative plasma biomarkers to aid in the diagnosis of VTE including selectins, microparticles, IL-10 and other inflammatory markers. These molecular markers may also predict recurrence risk, guide length and modality of treatment, and predict which thrombi will resolve spontaneously or recanalize, thus potentially identifying patients who would benefit from more aggressive therapies than standard anticoagulation.

Keywords::

• D-dimer
• deep venous thrombosis
• E-selectin
• microparticles
• P-selectin
• venous thromboembolism

1. Introduction

Primary and recurrent venous thromboembolism (VTE) persists as a source of major morbidity and mortality in hospitalized patients with > 900,000 primary and recurrent events occurring in 2002 [1]. As non-specific subjective complaints and a paucity of objective clinical examination findings complicate the diagnosis of VTE, diagnostic modalities serve an essential role. Compression ultrasound remains the gold standard for deep venous thrombosis (DVT) diagnosis. Reliable imaging is not always available, however, making a sensitive and specific serologic diagnosis, or biomarker, highly desirable. While D-dimer, a highly sensitive biomarker, is useful for excluding acute VTE, it lacks the specificity necessary for diagnostic confirmation. As such, ongoing research efforts target and support the utility of alternative plasma biomarkers for VTE diagnosis including selectins, microparticles (MPs), IL-10 and other cytokines. These molecular markers (biomarkers), while important in venous thrombogenesis (i.e., P-selectin), may also be useful to target the length of appropriate anticoagulation treatment and to predict thrombus biological activity; these molecular markers are reviewed here.

1.1 D-dimer

D-dimer, a derivative of cross-linked fibrin, has been well studied and previously linked with both DVT and pulmonary embolism (PE). The sensitivity of D-dimer for the diagnosis of DVT is reported as 96% for the diagnosis of DVT, but serves as a poor biomarker for DVT given its low specificity (40%) and low positive predictive value (PPV) (48%) [2,3]. However, a normal plasma D-dimer level serves as a test of exclusion for DVT. D-dimer levels have also been studied for use in guiding the length of therapy for initial VTE in several prospective trials [4,5]. It has been reported that abnormal D-dimer levels at 1 month following anticoagulation therapy withdrawal are an independent risk factor for recurrent VTE, and some authors support a clear benefit of prolonged therapy with Vitamin K antagonists in patients whose D-dimer levels are abnormal 1 month following discontinuation of anticoagulation. Finally, D-dimer levels have been correlated with estimated thrombus volume [6].

1.2 P-selectin

P-selectin is an adhesion glycoprotein present in platelet α-granules and endothelial cell Weibel–Palade bodies [7]. P-selectin translocates to the cell surface following exposure to an activating stimulus (i.e., thrombin or histamine) and in this ligand–receptive form binds to P-selectin glycoprotein ligand 1 (PSGL-1), a homodimeric mucin found on the majority of leukocytes and in small amounts on platelets [8,9]. This binding initiates a signaling pathway that mediates leukocyte–endothelial cell, leukocyte–platelet, leukocyte–leukocyte and platelet–endothelial interactions and thus leukocyte rolling and the initiation of an inflammatory cascade [10]. This interaction also releases procoagulant MPs, which carry tissue factor (TF), the initial trigger for thrombogenesis and other procoagulant factors [11]. P-selectin induces the exposure of phosphatidylserine and up-regulation of TF on monocytes, modulating initial thrombus amplification [12]. A soluble form of P-selectin (sP-sel) exists in the circulation, either originating from an alternatively spliced form found in platelets and endothelial cells (lacking a transmembrane domain) or reflecting proteolytic cleavage from the cell membrane shortly after activation [13]. Growing evidence supports that platelet activation increases levels of sP-sel [14]. While it has been theorized that plasma levels of sP-sel may predict thrombotic consumptive platelet disorders, these levels may also reflect endothelial cell activation or function in an anti-inflammatory role in which sP-sel directly binds PSGL-1 on leukocytes thereby preventing leukocyte–vessel wall interactions and reducing their ability to adhere to endothelium in vivo [15]. Mouse studies have previously supported a role for P-selectin in hemostasis [16].

P-selectin has been found to be intimately associated with venous thrombogenesis experimentally as the inhibition of P-selectin causes an increase in thrombus regression and vein recanalization and a decrease in thrombus formation and vein wall fibrosis [17]. It is also clear that plasma levels of sP-sel are elevated in acute DVT [2,18]. Rectenwald et al. compared P-selectin levels in 21 patients with DVT with 30 healthy controls and found them to be elevated at 88.7 ng/ml (compared with 22.1 ng/ml) [2]. Elevated levels of sP-sel have been associated with an increased risk of recurrent DVT and also found to be predictive of VTE in cancer patients [18-20]. Gremmel et al. demonstrated that while statistically significantly elevated at the time of DVT diagnosis compared with healthy controls, sP-sel and D-dimer levels decrease to a level similar to controls if treated with oral anticoagulation at 12 months [18]. On the contrary, however, these markers increase significantly following anticoagulation cessation at 6 months. Finally, Ramacciotti et al. used a logistic regression model to analyze 178 patients presenting for duplex imaging to diagnose DVT [11]. While sP-sel exhibited excellent specificity (96%), the addition of a Wells' score increased the PPV of P-selectin from 72 to 100%. As such, a sP-sel level greater than 90 ng/ml combined with a Wells' score 2 or greater was found to be as effective as duplex ultrasound at ruling in the diagnosis of DVT, while a sP-sel level less than 60 ng/ml with a Wells' score less than 2 was able to effectively ‘rule-out' the diagnosis (Table 1). A similar study from the same institution performed to improve the statistical validity of the aforementioned findings used a different validation cohort to analyze the role of sP-sel in the diagnosis DVT [21]. One hundred and fifty-nine patients were analyzed including 80 patients with confirmed DVT and 79 controls. Again, patients with positive DVT compared with those without DVT had statistically significantly higher P-selectin; D-dimer, C-reactive protein (CRP) and von Willebrand factor levels were also measured and elevated. Specifically, sP-sel levels > 90 ng/ml in this study combined with a Wells' score ≥ 2 or greater offered a 96% specificity and 89% PPV in diagnosing acute lower extremity DVT. While sP-sel proved more effective in ruling-in the diagnosis of DVT, the negative predictive value (NPV) cited in the current study was only 77% and thus much less effective than D-dimer in excluding the diagnosis of DVT, as the use of D-dimer < 500 ng/ml combined with a Wells' score < 2 demonstrated a 100% NPV.

Table 1. Summary of p value, sensitivity and specificity, PPV and NPV derived from a multivariable logistic regression model in which two combined biomarkers and/or a biomarker plus clinical score is the independent variable.

Variables	p-Value (regression)	Sensitivity (%)	Specificity (%)	NPV (%)	PPV (%)
sP-sel (≥ 90 ng/ml)	< 0.0001	28	96	72	72
sP-sel + Wells' score (≥ 90 ng/ml + ≥ 2)	< 0.0001	33	95	70	100
sP-sel + Wells' score (< 60 ng/ml + < 2)	< 0.0001	99	33	96	47
D-dimer (≤ 0.5 mg/l)	< 0.0001	98	29	80	40
D-dimer + Wells' score (≤ 0.5 mg/l + < 2)	< 0.0001	93	45	81	44
Wells' score (≥ 2)	< 0.0001	41	93	79	33
sP-sel + D-dimer (≥ 90 ng/ml + ≤ 0.5 mg/l)	< 0.0001	43	81	81	58

Adapted from [11] with permission of Sage Publications.

DVT: Deep venous thrombosis; NPV: Negative predictive value; PPV: Positive predictive value; sP-sel: Soluble P-selectin.

1.3 E-selectin

E-selectin temporally follows P-selectin up-regulation and has been noted to augment the thrombotic response in a murine model of venous thrombosis and to amplify the effects of P-selectin [22]. Additionally, homozygosity of the single nucleotide polymorphism Ser128Arg in the E-selectin gene in humans may be associated with recurrent VTE [23]. In the author's laboratory, E-selectin inhibition combined with P-selectin inhibition has been found an effective strategy to decrease venous thrombosis.

1.4 Microparticles

MPs are small fragments of cell membrane phospholipids, < 1 micron in size, that are shed in response to cell activation, injury or apoptosis from platelets, leukocytes and endothelial cells [2]. Rich in TF, phosphatidylserine and PSGL-1 MPs facilitate and amplify coagulation when recruited into a developing thrombus [2,24]. MPs are prothrombotic as they support the assembly of the prothrombinase complex on their membrane and monocyte-derived MPs activate endothelial cells in vitro leading to the expression of TF [16]. In early venous thrombogenesis, MPs are recruited to the area of thrombosis, where they amplify coagulation via TF and factor VIIa [25]. MPs play a central role in both the coagulation and inflammatory processes at play during acute thrombogenesis. Elevated levels of MPs have been recognized in a murine venous thrombosis model [26]. Elevated MP–monocyte conjugates, and increased platelet activation have been demonstrated in patients with VTE when compared with healthy controls supporting the role of elevated MPs and their binding to monocytes as key events in thrombogenesis [27]. Additionally, high levels of circulating MPs may serve as an independent risk factor for VTE [28]. Recently, MPs were extracted from platelet-poor plasma from nine patients with documented DVT by duplex ultrasound and compared with healthy controls [29]. Proteomic analysis was performed for each sample. Two proteins (galectin-3 binding protein (Gal3BP) and alpha-2 macroglobulin (A2M)) were differentially expressed on DVT patients. Gal3BP precursor is a polypeptide from the lectin family that promotes integrin-mediated cell adhesion; this protein plays a key role in platelet activation and function, synergizes with adenosine diphosphate (ADP) or thrombin to induce platelet aggregation and adenosine triphosphate (ATP) release, induces up-regulation of P-selectin, induces GPIIIa expression, promotes shedding of MPs and favors the generation of leukocyte–platelet aggregates supporting its important role in thrombosis and inflammation, potentially amplifying the thrombus progression. Nine proteins were depleted including fibrinogen β- and γ-chain precursors. The current difficulty with MPs as a diagnostic tool is their difficulty in measurement and poor reproducibility along with the lack of protocol standardization with no literature consensus on their measurement.

1.5 Additional molecular markers

Thrombin generation plays an integral role in hemostasis, and there exist certain risk factors for thrombin generation that are also associated with recurrent VTE [30]. Thrombin as a molecular marker to stratify DVT recurrence risk has been reported. IL-10, a major immunoregulatory and anti-inflammatory cytokine has been linked to DVT. Animal models have demonstrated that IL-10 modulates the development of DVT [31]. Additionally, a human case–control study evaluated IL-10 polymorphisms as a risk factor for venous thrombosis; the authors demonstrated that IL-10 G13 and G10 alleles were independent risk factors for venous thrombosis and that the IL-10 G10 allele was more frequent in recurrent disease, suggesting that it may be an independent risk factor for recurrent disease.

2. Conclusion

While the traditional diagnosis of VTE relies on primarily imaging studies, a multimodality approach including serologic testing continues to evolve. Ongoing data support the use of molecular markers (biomarkers) that may not only aid in the diagnosis of VTE, but also predict recurrence risk and guide length and modality of treatment. D-dimer, sP-sel, MPs and other markers have the potential to aid in the diagnosis of VTE (when other imaging modalities are not available), determine length of anticoagulation necessary for an individual patient and help predict which thrombi will resolve spontaneously or recanalize, and those that will not. Thus, inclusion of biomarkers may potentially identify patients who would benefit from more aggressive therapies than standard anticoagulation.

3. Expert opinion

As aforementioned, extensive ongoing research efforts appear to support the utility of alternative plasma biomarkers for the diagnosis of acute VTE. To date, it's been shown that D-dimer effectively serves as a test of exclusion for DVT, especially when combined with a clinical risk prediction such as the Wells' score. Additionally, abnormal D-dimer levels 1 month after the conclusion of anticoagulant therapy can predict VTE recurrence and thus, help dictate continuation and duration of therapy. sP-sel is clearly elevated in the setting of acute VTE, and when measured in combination with a Wells' score, demonstrates excellent specificity in diagnosing acute lower extremity DVT. While research supporting the role of E-selectin, MPs, thrombin and IL-10 is promising, ongoing efforts and results will be required before these biomarkers become clinically useful. Inhibition of E-selectin and P-selectin for either prophylaxis or therapy warrants ongoing research efforts. Additionally, if MPs are to be of clinical utility in addition to being a research tool, standardization, reproducibility and ease of measurement will need to be improved.

This field of research holds significant potential. These biomarkers hold promise to both explain the pathogenesis of thrombosis and serve as a marker for the thrombotic process. For example, while sP-sel is a good biomarker for thrombosis, laboratory data suggest that P-selectin is intimately involved in the thrombogenic process. Additionally, not only could a simple serum- or plasma-based assay(s) diagnose acute VTE in the future, especially in the absence of available efficient duplex imaging, but DVT-specific biomarkers might also aid in the estimation of the extent or spontaneous resolution of thrombus burden, predict recurrence and/or stratify the highest risk patients to more aggressive anticoagulation therapies. Such biomarkers may potentially serve as future targets for therapy. While this field of research is promising, these new tests are preliminary and require in most cases confirmation with large observational studies as well as laboratory standardization before they can be utilized in clinical practice. Additionally, before widespread utilization is possible, simplified techniques for these assays require development. For this field to move forward, additional translational research is needed. By first attempting to use these biomarkers in experimental models to aid in understanding the pathophysiology and clinical diagnosis of DVT, and to characterize thrombus extent, new agents can then be fashioned to target these same molecules for the prophylaxis and treatment of venous thrombosis. Ultimately, should these model systems prove to be fruitful, they can be extended to human pathophysiology and clinical trials. Laboratory efforts focusing specifically on P-selectin and sP-sel have proved most encouraging. Finally, proteomic and metabolomic techniques may also move the field forward in the future.

Declaration of interest

The authors state no conflict of interest and have received no payment in preparation of this manuscript.