Thrombin Generation Markers as Predictors of Cancer-Associated Venous Thromboembolism: A Systematic Review

Abstract

Venous thromboembolism (VTE) is a main contributor to morbidity and mortality in cancer patients. Biomarkers with the potential to predict cancer-associated VTE are continually sought. Of these, markers of thrombin generation present a likely option. The present systematic review examines the ability of three widely used biomarkers of thrombin generation: prothrombin fragment 1.2 (F1.2), thrombin-antithrombin complex (TAT), and ex vivo thrombin generation, to predict VTE in both solid and hematologic adult cancer patients. Relevant studies were identified in the PubMed and Embase databases, and the review conformed to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines. Each study was evaluated using the quality assessment tool from the National Heart, Lung, and Blood Institute. The review protocol was published on PROSPERO with identifier CRD42022362339. In total, 24 papers were included in the review: 11 reporting data on F1.2, 9 on TAT, and 12 on ex vivo thrombin generation. The quality ratings of the included studies varied from good (n = 13), fair (n = 8), to poor (n = 3) with a high heterogenicity. However, F1.2, TAT complex, and ex vivo thrombin generation were all found to be associated with the development of VTE. This association was most pronounced for F1.2. Furthermore, the determination of F1.2 was able to improve the precision of several established risk assessment scores. In conclusion, markers of thrombin generation were found to be elevated in cancer patients with VTE, and particularly, F1.2 was found to be a promising predictor of cancer-associated VTE.

Keywords

prothrombin fragment 1.2 - antithrombin III-protease complex - thrombin generation - cancer-associated thrombosis - embolism and thrombosis - neoplasms

Thrombosis is the second-most common cause of death in cancer patients after the cancer itself and is a significant contributor to morbidity and decreased quality of life in these patients.[1] [2] [3] Both arterial thrombosis and venous thromboembolism (VTE) occur in cancer, and while arterial thrombosis, especially ischemic stroke,[4] in cancer has gained focus in recent years, VTE is still the most prevalent type of thrombosis in cancer. The present review focuses on the prediction of VTE in patients with cancer and does not cover arterial thrombosis.

VTE can be effectively prevented with thromboprophylactic drugs. However, these drugs potentiate the increased risk of bleeding already present in cancer patients.[5] [6] It is therefore important to correctly identify and select cancer patients who can benefit from thromboprophylaxis and avoid treating those where treatment is unnecessary or even harmful. To this end, several risk assessment models have been developed to help clinicians identify cancer patients at increased risk of VTE.[7] [8] [9] [10] [11]

The Khorana score forms the basis of most of the scores. It was developed in 2008 and designed for the prediction of VTE in cancer patients receiving chemotherapy.[7] Several later scores have expanded or amended the Khorana scores, such as the Vienna Cancer and Thrombosis (CATS) score,[8] the PROTECHT score,[9] and the ONKOTEV score.[10] Other scores have been developed independently of the Khorana score, aiming to reduce the complexity of the scoring and increase clinical applicability.[11] [12] The risk assessment models have been validated in several subsequent studies.[13] [14] [15] Here, the results have been mixed, and the clinical applicability of the scores has varied.[13] [15] All the risk scores rely on a combination of clinical data and biochemical parameters. An important aspect of the work to improve these scores is the investigation of additional potential biomarkers of VTE.

A promising category of biomarkers is analyses measuring thrombin activity. Thrombin is the product of the coagulation cascade, responsible for converting fibrinogen to fibrin and forming a stable clot. Apart from its role in clot formation, thrombin also inhibits fibrinolysis through thrombin activatable fibrinolysis inhibitor and activates platelets through their thrombin receptor.[16] It is well established that cancer patients have increased thrombin formation, driven by an increase in tissue factor expression on endothelium and circulating tumor-derived microparticles and through interactions with the endothelium, immune cells, and platelets.[16] [17] Anticancer therapies may further enhance thrombin formation ([Fig. 1]). Interestingly, thrombin also seems to promote cancer growth and development by interactions with the endothelium and the tumor cells.[16] [17] [18] As such, thrombin is an integral player in the development of thrombosis in cancer and might be a promising biomarker for VTE prediction. However, due to the short half-life of thrombin in blood, it is not possible to directly quantify the amount of circulatory thrombin at any given time.[16] Therefore, it is necessary to measure more stable by-products of thrombin (e.g., prothrombin fragment 1.2 (F1.2) and thrombin-antithrombin complex (TAT), or ex vivo thrombin generation potential). F1.2 is the by-product when thrombin is cleaved from prothrombin and is generated in a 1:1 ratio to thrombin.[19] TAT formation takes place when active thrombin is bound by antithrombin. TAT increases when there is an overflow of free thrombin and is therefore increased in the prothrombotic state.[16] [20] Both thus reflect in vivo thrombin formation, and both have considerably longer half-life than thrombin and are therefore better suited as biomarkers.

Fig. 1 Thrombin generation in cancer. The figure shows a simplified outline of some of the mechanisms behind increased thrombin generation in cancer. Cancer cells may directly express tissue factor as well as release negatively charged EVs whose surface promote coagulation. Cancer cells also increase inflammation in the form of proinflammatory cytokines that further thrombin generation. Lastly, cancer cells increase platelet and vascular endothelium activation, which also leads to thrombin generation. Surgery induces tissue damage that also furthers coagulation, and chemotherapy increases TF activity, activates platelets, and induces procoagulant apoptotic cell debris.[17] [18] EVs, extracellular vesicle(s); TF, tissue factor; VWF, von Willebrand Factor. Figure created with BioRender.

Thrombin generation can also be measured ex vivo by the addition of activators (calcium, phospholipids, and tissue factor) and a thrombin-specific, fluorogenic substrate to platelet-poor or -rich plasma. The resulting thrombin generation curve is conventionally described with the parameters lagtime, time to peak, peak thrombin concentration, and endogenous thrombin potential (ETP).[21] [22] Furthermore, the mean rate index may be calculated as peak thrombin concentration/(time to peak – lagtime).

F1.2, TAT, and ex vivo thrombin generation are all expressions of thrombin activity. However, their applicability as VTE biomarkers in a clinical setting is not sufficiently understood. We therefore aimed to systematically review the existing evidence in the literature to examine if these three selected markers of thrombin generation can predict VTE in cancer patients.

Materials and Methods

The present review follows the Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines.[23] The search was performed in the PubMed and Embase databases in July 2023. Search strings were composed of indexed terms and freeform terms for cancer, thrombosis and F1.2 and/or TAT and/or thrombin generation. The full search strings are presented in [Table 1].

Table 1 Search strings in PubMed and Embase

PubMed

(“Neoplasms”[MeSH Terms] OR cancer* OR neoplasm* OR malignanc*) AND (“Embolism and Thrombosis”[MeSH Terms] OR “venous thromboembolism” OR “venous thrombo-embolism” OR “venous thrombosis” OR “deep vein thromboembolism” OR “deep vein thrombosis” OR “portal thrombosis” OR “sinus thrombosis” OR “mesenterial thrombosis” OR “cerebral vein thrombosis” OR “cancer-associated thrombosis” OR “VTE” OR “DVT”) AND (“Thrombin”[MeSH] OR “thrombin” OR “calibrated automated thrombogram” OR “thrombography” OR “prothrombin fragment 1.2” [Supplementary Concept] OR “prothrombin” OR “antithrombin III-protease complex” [Supplementary Concept] OR “TAT” OR “thrombin-antithrombin”)

Embase

('malignant neoplasm'/exp OR cancer* OR neoplasm* OR malignanc*) AND ('thromboembolism'/exp OR “venous thromboembolism” OR “venous thrombo-embolism” OR “venous thrombosis” OR “deep vein thromboembolism” OR “deep vein thrombosis” OR “portal thrombosis” OR “sinus thrombosis” OR “mesenterial thrombosis” OR “cerebral vein thrombosis” OR “cancer-associated thrombosis” OR “VTE” OR “DVT”) AND ('thrombin'/exp OR 'thrombography'/exp OR “thrombin” OR “calibrated automated thrombogram” OR “thrombography” OR “prothrombin” OR 'thrombin antithrombin complex'/exp OR “TAT” OR “thrombin-antithrombin”) NOT 'conference abstract'/it

Inclusion Criteria

We included studies on patients with both solid and hematologic cancers that recorded VTE and included measurements on F1.2, TAT, and/or ex vivo thrombin generation. VTE was defined as either (1) distal or proximal deep venous thrombosis in upper or lower extremity, (2) pulmonary embolism, (3) cerebral venous sinus thrombosis, or (4) splanchnic, portal, or other thromboses in veins in or supplying organs and which had been confirmed with relevant imaging techniques. Prospective cohort, case-control, cross-sectional, post-hoc, and randomized controlled studies in English were accepted.

Exclusion Criteria

Exclusion criteria were conference abstracts, studies with a nonpredictive design (diagnosis of VTE prior to measurement of thrombin generation parameters), studies that did not report their thrombin generation results in numbers or graphically, studies that did not perform statistical analysis of the association between thrombin generation parameters and VTE, studies with no original data, case reports with <10 cases, and studies that had no VTE events.

Data Extraction

After duplicate sorting and exclusion, 50 abstracts were screened by two authors (T.G. and J.B.L.) to validate the inclusion and exclusion criteria. The remaining abstracts were subsequently screened by either author (T.G. or J.B.L.). If eligibility could not be determined from the abstract, the study record was included for full-text screening. After abstract screening, 10 study records were randomly selected for full-text reading by both T.G. and J.B.L. to revalidate the inclusion and exclusion criteria. The remaining study records were read in full text by T.G. or J.B.L. and either included or excluded according to the predefined criteria.

Quality Assessment

Quality assessment of the included studies was performed by the authors T.G. and J.B.L. using the applicable study quality assessment tool from the National Heart, Lung, and Blood Institute.[24]

Data Synthesis

According to the protocol, more than five studies had to report a mean value for the selected biomarker in both the VTE and non-VTE groups, in order to conduct a meta-analysis. Two studies reported a mean F1.2, four reported a TAT mean, and only one study reported a mean thrombin generation. Thus, a qualitative synthesis of the included studies was performed.

Results

From a total of 4,510 papers identified in the literature search, 24 papers were included in the final review ([Fig. 2]). Of these, 11 studies reported data on F1.2[15] [19] [25] [26] [27] [28] [29] [30] [31] [32] [33] ([Table 2]), nine studies reported data on TAT[25] [26] [27] [30] [33] [34] [35] [36] [37] ([Table 3]), and 12 studies reported data on thrombin generation[13] [21] [28] [29] [31] [38] [39] [40] [41] [42] [43] [44] ([Table 4]). Of the 24 included papers, 13 were rated as good, eight as fair, and three studies were rated as poor.

Table 2 Studies measuring prothrombin factor 1.2 (F1.2)

Author, year, quality assessment	Study design	Population	Thromboprophylaxis	Assay, sample timing	Outcomes, VTE definition and incidence	Findings
Ay et al 2009[19] Good	Prospective observational[a]	Mixed cancer patients (n = 821). Age: 62 y (53-68) Female: 45% Site of cancer: Brain (12%), breast (16%), lung (15%), stomach (4%), colorectal (13%), pancreas (6%), kidney (3%), prostate (12%), multiple myeloma (2%), lymphoma (11%) and other (6%)	Excluded If therapeutic. Short-term heparin prophylaxis allowed but NR	Enzygnost F1 + 2 At inclusion	Incident symptomatic VTE confirmed with imaging or autopsy VTE incidence: 8% (n = 62) in median 501 d FU PE (47%), lower extremity DVT (44%) and other (10%)	F1.2 ↑ in VTE vs. non-VTE: 310 (213–416) vs. 249 (190–353) pmol/L, p = 0.003 With 75% cut-off: 358 pmol/L Univariate analysis: HR = 2.2 (CI: 1.3–3.6), p <  0.003 Multivariate, adjusted analysis: HR = 2.0 (CI: 1.2–3.6), p = 0.02
Falanga et al 1993[25] Fair	Prospective observational	Solid cancer patients undergoing surgery (n = 117) Age range: 40–62 y (median NR) Female: 36% Site of cancer: Gastrointestinal (80%), lung, breast, kidney, or thyroid (20%)	Heparin prophylaxis for 7 d (n = 83) No prophylaxis (n = 34)	Enzygnost F1 + 2 Before surgery.	Screening for incident VTE with 125I-fibrinogen uptake test after surgery, confirmed with imaging VTE incidence: 8% (n = 9) in 7 d FU All lower extremity DVT	F1.2 ↔ in VTE vs. non-VTE: Relative risk of VTE with cut-of ≥ 1.1 nM: RR = 1.75 (CI: 0.46–6.61), p NS, value NR
Iversen et al 1996[26] Good	Prospective observational[b]	Colorectal cancer patients undergoing surgery (n = 93). Age: 72 y (41–91) Female: 47%	Heparin prophylaxis for at least 7 d	Enzygnost F1 + 2 Before surgery	Screening for incident VTE with imaging after surgery VTE incidence: 22% in 12 d FU (n = 20) PE (5%), DVT (95%)	F1.2 ↔ in VTE vs. non-VTE: 1.8 nM (1.7–2.4) vs. 1.8 (1.6–2–0), p NS, value NR
Iversen et al 2002[27] Good	Prospective observational[b]	Colorectal cancer patients undergoing surgery (n = 137) Age: 75 y, range 41–84 Female: 50%	Heparin prophylaxis for at least 7 d	Enzygnost F1 + 2 Before surgery and postoperative day 1, 2, and 7	Screening for incident VTE with imaging after surgery in 93% of patients (n = 128) VTE incidence: 20% (n = 26) in 12 d FU PE + DVT (4%), DVT (96%)	F1.2 ↑ VTE vs. non-VTE[c]: Before surgery: Geometric mean 2.01 nmol/L (CI: 1.72–2.36) vs. 1.75 nmol/L (1.62–1.90), p NR Postoperative day 1: 3.80 vs. 2.15 nmol/L, p <  0.002 Postoperative day 2: 2.75 vs. 2.25 nmol/L, p <  0.01 Postoperative day 7: 3.75 vs. 3.35 nmol/L, p NR
Pépin et al 2016[15] Good	Case-control (derived from prospective observational cohort)	Ambulatory solid cancer patients receiving chemotherapy (n = 160, consisting of 20 cases with VTE and 140 controls without) Mean age: 61 ± 9.9 y Female: 40% Site of cancer: Pancreas (40%), gastrointestinal (40%), lung (15%), other (5%)	Patients in therapeutic doses excluded. Prophylaxis NR.	Enzygnost F1 + 2 At inclusion	Screening for incident VTE with imaging at inclusion. Further scanning if clinical suspicion VTE incidence in prospective cohort: 5% in 6 mo FU (n = 23 of 443) Type NR.	F1.2 ↔ VTE vs. non-VTE: Mean 439 ± 223 vs. 425 ± 781 pmol/L, p NS, value NR
Posch et al 2016[28] Good	Prospective observational[a]	Newly diagnosed or remitting cancer patients (n = 804) Age: 63 y (54–69) Female: 54% Site of cancer: Brain (11%), breast (17%), gastrointestinal (18%), kidney (3%), lung (15%), lymphoma (11%), other (8%), pancreas (7%), prostate (12%)	Anticoagulants not allowed	Enzygnost F1 + 2 At inclusion	Incident VTE detected by routine imaging or due to clinical suspicion VTE incidence: 7% (n = 55) in 2 y FU Type NR	F1.2 ↑ in VTE vs. non-VTE: Univariate risk score for VTE with 1 SD increase in F1 + 2: Subdistribution HR = 1.48 (CI: 1.23–1.78), p <  0.001
Reitter et al 2016[29] Good	Prospective observational[a]	Solid cancer patients beginning chemo- or radiotherapy (n = 112) Age: 62 y, range 21–80 Female: 43% Site of cancer: Brain (35%), colorectal (13%), lung (37%) and pancreas (15%)	Short-term heparin prophylaxis allowed but NR	Enzygnost F1 + 2 At inclusion	Symptomatic incident VTE confirmed by imaging or autopsy report VTE incidence: 13% (n = 14) in 250 d FU PE (29%), DVT (50%) and other	F1.2 ↑ in VTE vs. non-VTE: Per twofold increase in F1 + 2 HR  =  2.11 (CI: 1.47–3.03), p <  0.0001
Sallah et al 2004[30] Good	Prospective observational	Solid cancer patients with diagnosis of first VTE (n = 223) Age: 57 y (IQR NR) Female: 50% Site of cancer: Breast (20%), colorectal (16%), genitourinary (17%), head and neck (14%), lung (18%), and other (15%)	All patients on vitamin K antagonists for at least 6 mo	Enzygnost F1 + 2 Before initiation of anticoagulation	VTE recurrence confirmed by imaging Incidence of recurrent VTE: 14% (n = 31) Type NR	F1.2 ↔ in VTE vs. non-VTE: Full cohort: Median 3.7 and mean 4.3 nmol/L VTE recurrence group: Median 4.4 and mean 5.1 nmol/L, p and IQR/range NR Univariate regression for VTE recurrence (cut-off NR): OR = 1.30 (CI: 0.97–1.73), p = 0.07
Thaler et al 2014[31] Good	Prospective observational[a]	Newly diagnosed high-grade glioma patients, 2–3 wk after surgery (n = 141) Age: 54 y (44–64) Female: 33%	Patients in therapeutic doses excluded. Prophylaxis 8–10 d postsurgery	Enzygnost F1 + 2 Sampling at inclusion	Symptomatic incident VTE confirmed by imaging or autopsy report VTE incidence: 17% in 2 y FU (n = 24) PE (46%), DVT (54%)	F1.2 ↔ VTE vs. whole cohort: 286 pmol/L (167–435) vs. 224 pmol/L (156–324), p NR Univariate HR for prediction of VTE per doubling of F1.2: 1.59 (CI: 1.02–2.41), p = 0.041 Multivariate HR = 1.53 (0.97–2.36), p = 0.069
Tsubata et al 2022[32] Poor	Prospective observational	Nontreatable lung cancer (newly diagnosed or remission) (n = 1,008) Age: 70 y, range 30–94 Female: 29%	NR. Some patients receiving Edoxaban treatment included in this cohort	Assay NR Timing NR	All patients screened with CT and ultrasound at inclusion. Symptomatic incident and recurrent VTE were recorded during the FU period. VTE incidence: 4% in 2 y FU (n = 38) Type NR	F1.2 ↑ in VTE vs. non-VTE: Univariate analysis of VTE risk with cut-off F1 + 2 ≥ 325 pmol/L OR = 1.120 (CI: 1.073–1.169), p = 0.000 Risk assessment model for the development of VTE: Coefficient = 0.768, OR = 2.155 (CI: 1.313–3.535), p = 0.002
van Doormaal et al 2012[33] Fair	Prospective observational	Mixed cancer patients (n = 43) Age: 59 ± 12 y Female: 42% Site of cancer: Breast (19%), gastrointestinal (37%), pancreatic (30%) and other (14%)	No anticoagulant use allowed at inclusion	Enzygnost F1 + 2 Sampling at inclusion	Symptomatic incident VTE confirmed with imaging and recorded during the FU period VTE-incidence: 11.6% (n = 5) in 6 mo FU PE (40%), DVT (40%) and other (20%)	F1.2 ↔ VTE vs. non-VTE group: 319 (187–558) vs. 241 (179–312) pmol/L, p = 0.427

Abbreviations in table: AUC, area under the curve, CI, 95% confidence interval, CT, computed tomography, DVT, deep venous thrombosis, ELISA, enzyme-linked immunosorbent assay, F1 + 2, prothrombin fragment 1 + 2, FU, follow-up, HR, hazard ratio, IQR, interquartile range, LMWH, low molecular weight heparin, MRI, magnetic resonance imaging, NR, not reported, NS, nonsignificant, OR, odds ratio, ROC, receiver operating characteristic, RR, relative risk, SD, standard deviation, SE, standard error, TF, tissue factor, VTE, venous thromboembolic event(s).

Note: Age and findings reported as median (IQR) unless stated otherwise.

Note: ↔ no change, ↑ increased.

^a Study population part of the same cohort (the Vienna CATS-study).

^b Study population part of the same cohort.

^c Findings on postoperative day 1, 2 and 7 were read from Figure 3 in publication.

Table 3 Studies measuring thrombin-antithrombin complex (TAT)

Author, year, quality assessment	Study design	Population	Thromboprophylaxis	Assay, sample timing	Outcomes, VTE definition and incidence	Findings
Falanga et al 1993[25] Fair	Prospective observational	Solid cancer patients undergoing surgery (n = 117). Age range: 40–62 y (median NR) Female: 36% Site of cancer: Gastrointestinal (80%), lung, breast, kidney or thyroid (20%)	Heparin prophylaxis for 7 d (n = 83) No prophylaxis (n = 34)	Enzygnost TAT Before surgery	Screening for incident VTE with 125I-fibrinogen uptake test after surgery, confirmed with imaging VTE incidence: 8% (n = 9) in 7 d FU All lower extremity DVT	TAT ↑ in VTE vs. non-VTE: Risk of with VTE with cut-off > 3.5 ng/mL: RR = 7.5 (CI: 1.65–34.6), p <  0.01. Sensitivity 78%, specificity 72%, positive predictive value 23% and negative predictive value 97%.
Iversen et al 1996[26] Good	Prospective observationalb	Colorectal cancer patients undergoing surgery (n = 93) Age: 72 y (41–91) Female: 47%	Heparin prophylaxis for at least 7 d	Enzygnost TAT Before surgery	Screening for incident VTE with imaging after surgery VTE incidence: 22% (n = 20) PE (5%), DVT (95%).	TAT ↔ in VTE vs. non-VTE: 3.5 (CI: 3.0–6.0) vs. 3.0 µg/L (3.0–4.0), p NS, value NR
Iversen and Thorlacius-Ussing 2002[27] Good	Prospective observationalb	Colorectal cancer patients undergoing surgery (n = 137) Age: 75 y, range 41–84 Female: 50%	Heparin prophylaxis for at least 7 d	Enzygnost TAT Before surgery	Screening for incident VTE with imaging after surgery in 93% of patients (n = 128) VTE incidence: 20% (n = 26) in 12 d FU PE + DVT (4%), DVT (96%)	TAT ↑ VTE vs. non-VTE[c]: Before surgery: Geometric mean: 3.83 (CI: 3.07–4.78) vs. 3.72 μg/L (3.30–4.19), p NR Postoperative day 1: 15 vs. 12 μg/L, p <  0.03 Postoperative day 2: 11 vs. 9.5 μg/L, p NR Postoperative day 7: 9 vs. 7 μg/L, p NR
Kirwan et al 2008[34] Poor	Prospective observational	Breast cancer patients commencing chemotherapy as an incidence (postsurgery) treatment or as treatment for metastatic disease (n = 123) Age: 52 y, range 31–78 Female: 100%	No anticoagulant use allowed at inclusion	Enzygnost TAT Before chemotherapy	Screening for incident VTE with ultrasound after 1 month and if symptoms VTE incidence: 10% (n = 12) within 3 mo FU Type NR	TAT (↑) in VTE vs. non-VTE: Mean 8.9 (CI: 2.4–32.5) vs. 5.7 µg/mL (4.0–8.0), p NS, value NR
Liang et al 2020[35] Fair	Prospective observational	Cervical cancer patients admitted to hospital for treatment (n = 134) Age: 51 y, range 30–73 Female: 100%	No anticoagulant use allowed at inclusion	HISCL5000 First day of admission	Examination for incident VTE with ultrasound Timing NR VTE incidence: 7% (n = 10) Type NR	TAT ↑ in VTE vs. non-VTE: 3.28 ± 0.26 vs. 1.71 ± 0.14 ng/mL, (NR if means or medians), p = 0.004
Qi et al 2020[36] Poor	Observational retrospective	Solid cancer patients undergoing surgery (n = 286) Age, VTE: 62 ± 5 y, non-VTE: 61 ± 5 y Female: 45%	Patients in therapeutic doses excluded. Prophylaxis NR	Assay NR. During hospital stay	Screening for incident VTE with ultrasound Timing NR 22% (n = 63) VTE in 10 mo FU All lower extremity DVT	TAT ↑ in VTE vs. non-VTE: Mean 4.67 (SD ± 1.13) vs. 2.08 (± 0.75) µg/L, p = 0.036 In logistic regression per µg/L increase, RR = 6.122 (CI: 2.244–16.695) ROC-curve AUC = 0.698, cutoff value 4.58 µg/L
Sallah et al 2004[30] Good	Prospective observational	Solid cancer patients with diagnosis of first VTE (n = 223) Age: 57 y (IQR NR) Female: 50% Site of cancer: Breast (20%), colorectal (16%), genitourinary (17%), head and neck (14%), lung (18%) and other (15%)	All patients on VKA for at least 6 mo	TAT Antithrombin (Enzyme Research, Inc., South Bend, IN) Before initiation of anticoagulation	VTE was an inclusion criterion for the study. VTE recurrence confirmed by imaging Incidence of recurrent VTE: 14% (n = 31) Type NR	TAT ↑ in VTE vs. non-VTE: Full cohort: Median 5.3 and mean 6.1 ng/mL VTE recurrence cohort: Median 8.4 and mean 8.4 ng/mL Univariate regression for VTE recurrence (cut-off NR): OR = 1.57 (CI: 1.09–2.27), p = 0.01
Takata et al 2021[37] Good	Retrospective observational	Hepatocellular carcinoma patients undergoing hepatectomy (n = 65) Mean age: 74 y ± 2 y Female: 15%	Preoperative antithrombotic treatment (type NR) 14% of cohort (n = 9)	TAT Immunoassay (LSI Medience Corporation, Tokyo, Japan) Before surgery	Screening for portal vein thrombosis with CT before and 7 d after surgery Portal vein thrombosis incidence: 20% (n = 13) in 7 d FU	TAT (↑) VTE vs. non-VTE[c]: Before surgery: Mean 1.8 (SE ± 0.2) vs. 3.0 ng/mL (±0.4), p = 0.14. Postoperative day 1: 20.2 ± 3.1 vs. 15.7 ± 1.6 ng/mL, p = 0.11 Postoperative day 2: 12.6 ± 1.8 vs. 11.9 ± 1.0 ng/mL, p = 0.63 Postoperative day 3: 11.4 ± 2.2 vs. 10.7 ± 0.8 ng/mL, p = 0.86 Risk of VTE based on ratio of TAT postoperative day 1 vs. preoperative, cut-off NR: OR = 1.20 (CI: 1.01–1.42), p <  0.05. ROC-curve on postoperative day 1 with cutoff 5.73 ng/mL: AUC = 0.786, p <  0.001
van Doormaal et al 2012[33] Fair	Prospective observational	Mixed cancer patients (n = 43) Age: 59 y ± 12 y Female: 42% Site of cancer: Breast (19%), gastrointestinal (37%), pancreatic (30%) and other (14%)	No anticoagulant use allowed at inclusion.	Enzygnost TAT At inclusion	Symptomatic incident VTE confirmed with imaging during the FU period VTE-incidence: 11.6% (n = 5) in 6 mo FU PE (40%), DVT (40%) and other (20%)	TAT ↔ in VTE vs. non-VTE: 5.7 (IQR: 3.5–13.7) vs. 4.1 mg/L (3.2–6.2), p = 0.29

Abbreviations: AUC, area under the curve; CI, 95% confidence interval; CT, computed tomography; DVT, deep venous thrombosis; ELISA, enzyme-linked immunosorbent assay; FU, follow-up; HR, hazard ratio; IQR, interquartile range; LMWH, low molecular weight heparin; MRI, magnetic resonance imaging; NR, not reported; NS, nonsignificant, OR, odds ratio, ROC, receiver operating characteristic, RR, relative risk, SD, standard deviation, SE, standard error, TAT, thrombin-antithrombin complex, TF, tissue factor, VTE, venous thromboembolic event(s).

Note: ↔ no change, ↑ increased, (↑) discreetly increased.

Note: Age and findings reported as median (IQR) unless stated otherwise.

^a Study population part of the same cohort (the Vienna CATS-study).

^b Study population part of the same cohort.

^c Findings on postoperative day 1, 2, and 7 were read from Fig. 3 in publication.

Table 4 Studies measuring ex vivo thrombin generation

Author, year, quality assessment	Study design	Population	Thromboprophylaxis	Assay, sample timing	Outcomes, VTE definition and incidence	Findings
Abu Saadeh et al 2016[38] Fair	Prospective observational	Ovarian or endometrial adenocarcinoma patients undergoing surgery (n = 96). Age, VTE: 60 y (55–66) Non-VTE: 58 (48–66) Female: 100%	LMWH (Tinzaparin) 75 IU/kg daily for duration of hospital stay	Thrombinoscope Before surgery	Incident symptomatic VTE, verified with imaging VTE incidence: 8% (n = 8) in 180 d FU PE (n = 3) and lower extremity DVT (n = 5)	TG (↑) in VTE (n = 5) vs. non-VTE (n = 43): ETP: 1,928.2 (421.1) vs. 1,622.6 (475) nM x min, p <  0.05 Peak: 327.7 (73.1) vs. 288.8 (87.3) nM, p NR ttPeak: 5.7 (1.9) vs. 6.57 (1.72) min, p NR Lagtime: 3.01 (1.47) vs. 3.78 (1.6) min, p NR
Ay et al 2011[21] Good	Prospective observational[a]	Mixed cancer patients (n = 1,033). Age: 62 y (53–68). Female: 55% Site of cancer: Brain (13%), breast (15%), lung (14%), stomach (4%), colorectal (12%), pancreas (7%), kidney (3%), prostate (12%), multiple myeloma (3%), lymphoma (12%) and other (5%)	Short-term heparin prophylaxis allowed	Technothrombin At inclusion	Incident symptomatic VTE, confirmed with imaging or autopsy VTE incidence: 8% (n = 77) in 2 y FU PE (43%), lower extremity DVT (13%) and other (44%)	TG ↑ in VTE vs. non-VTE: ETP: 4.475 (IQR 4.087–4.915) vs. 4.386 (3.804–4.890) nM, p = 0.197 Peak: 556 (432–677) vs. 499 (360–603) nM, p = 0.14 ttPeak: 11.5 (9.5–14) vs. 13 (11–16) min, p = 0.006 Lagtime: 8 (7–10) vs. 9 (7–11) min, p = 0.036 Velocity index: 160 (111–223) vs. 127 (71–183) nM/min, p = 0.004 Cumulative risk of VTE in 6 mo for peak  <  611 nM: 11% vs. 4%, p = 0.002 Multivariate, adjusted analysis, per 100 nM increase in peak: HR = 1.15 (CI: 1.02–1.29), p = 0.019 Multivariate, adjusted analysis with 75th percentile cut-off (peak of 611 nM): HR = 1.9 (1.2–3.0), p = .008
Chalayer et al 2019[39] Good	Prospective observational	Newly diagnosed multiple myeloma patients undergoing first three cycles of chemotherapy (n = 71). Age: 67 y (59–73) Female: 55%	Short-term heparin prophylaxis and fondaparinux allowed	Thrombinoscope Before initiation of chemotherapy	Incident symptomatic arterial and VTE recorded at FU VTE incidence: 11% (n = 8) in median 47 d PE (25%), lower extremity DVT (62%), and other (13%)	TG ↔ in VTE vs. non-VTE: Peak: 186 (120–218) vs. 149 (114–181) nmol/L, p = 0.22 ttPeak: 10.8 (7.3–11.3) vs. 9 (8.4–10.2) min, p = 0.82
Gezelius et al 2018[40] Fair	Post-hoc of randomized controlled trial	Newly diagnosed small cell lung cancer patients undergoing chemotherapy (n = 242) Mean age, intervention group: 66 y ± 8 Control group: 67 ± 9 Female: 57%	Intervention group (n = 115): LMWH (Enoxaparin) 1 mg/kg for duration of chemotherapy Control group (n = 127): No prophylaxis	Thrombinoscope Before initiation of chemotherapy	Clinical examination for incident VTE at 2-month intervals VTE incidence, intervention group: 3% (n = 3) VTE incidence, control group: 9% (n = 12) Median FU time 276 days Type NR	TG ↔ in VTE vs. non-VTE: Peak: 236 (176–277) vs. 217 (176–261) nM, p = 0.42 ttPeak: 9.1 (7.4–11.2) vs. 10.0 (8.4–11.9) min, p = 0.26 ETP: 1,336 (1,164–1,471) vs. 1,222 (1,056–1,403) nM x min, p = 0.26
Leiba et al 2017[41] Fair	Prospective observational	Multiple myeloma patients (n = 33). Mean age: 63 y ± 9 Female: 47% 3 patients excluded from analysis due to VTE before sampling	Warfarin (n = 1, dose NR)	Thrombinoscope Before therapy start	Imaging performed if clinical suspicion of VTE VTE incidence: 24% (n = 8), mean FU 2.5 y PE (25%), DVT (38%) and other (38%)	TG ↑ in VTE vs. non-VTE: ETP: 2,896 vs. 2,028 nM x min, p <  0.001 Peak: 620 vs. 400 nM, p <  0.001 CI: NR
Posch et al 2016[28] Good	Prospective observational[a]	Newly diagnosed or remitting cancer patients (n = 804) Age: 63 y (54–69) Female: 54% Site of cancer: Brain (11%), breast (17%), gastrointestinal (18%), kidney (3%), lung (15%), lymphoma (11%), other (8%), pancreas (7%), prostate (12%).	Anticoagulants not allowed	Technothrombin At inclusion	Incident VTE detected by routine imaging or due to clinical suspicion VTE incidence: 7% (n = 55) in 2 y FU Type NR	TG ↑ in VTE vs. non-VTE: Univariate risk score for VTE with 1 SD increase in peak: Subdistribution HR = 1.46 (CI: 1.08–1.87), p = 0.01
Reitter et al 2016[29] Good	Prospective observational[a]	Solid cancer patients beginning chemo- or radiotherapy (n = 112) Age: 62 y, range 21–80 Female: 43% Site of cancer: Brain (35%), colorectal (13%), lung (37%) and pancreas (15%)	Short-term heparin prophylaxis allowed but NR	Technothrombin At inclusion	Symptomatic VTE recorded and confirmed by imaging or autopsy report VTE incidence: 13% (n = 14) in 250 days FU PE (29%), DVT (50%) and other	TG ↔ in VTE vs. non-VTE: HR per twofold increase in peak: 1.00 (CI: 0.99–1.00), p = 0.60
Schorling et al 2020[13] Fair	Prospective observational	Ambulatory solid cancer patients starting chemotherapy (n = 100) Age: 58 y (51–68) Female: 62% Site of cancer: Breast (30%), gastrointestinal (28%), genitourinary (16%), lung (4%), pancreas (6%) and other (16%)	Anticoagulants not allowed	Thrombinoscope Before therapy start	Symptomatic incident VTE and VTE discovered on routine imaging VTE-incidence: 11% (n = 10) in 3 mo FU PE + DVT (10%), PE (30%), DVT (50%), and other (10%)	TG ↔ in VTE vs. non-VTE: VTE group: Lagtime: 3.2 (1.8–5.8) ttPeak: 6.7 min (4.1–10.4) Peak: 260 nM (113–328) ETP: 1,689 nM x min (1,240–1,860) Non-VTE group Peak: 379.9 nM, p = 0.005 ETP:  <  than VTE group, value NR, p = 0.036 Full cohort: Lagtime: 2.2 min (1.5–3.0) ttPeak: 4.5 min (3.7–5.7) Peak: 364 nM (275–447) ETP: 1,917 nM x min (1,659–2,354)
Syrigos et al 2018[42] Good	Prospective observational	Ambulatory lung cancer patients (n = 150). Mean age: 65 y ± 10 Female: 27%	Anticoagulants not allowed	Thrombinoscope At inclusion	Symptomatic VTE, confirmed with imaging VTE-incidence: 8% (n = 12) at 6 mo FU PE (33%), DVT (58%) and other (8%)	TG ↑ in VTE vs. non-VTE: OR (CI) for development of VTE (cut-offs NR): Lagtime = 0.603 (0.294–1.235), p = 0.09 ttPeak = 0.522 (0.307–0.887), p = 0.09 Mean Rate Index = 1.016 (1.003–1.032), p = 0.02 Peak = 1.006 (0.997–1.016), p = 0.09 ETP = 1 (0.998–1.001), p = 0.1
Thaler et al 2014[31] Good	Prospective observational[a]	Newly diagnosed high-grade glioma patients, 2–3 wk after surgery (n = 141) Age: 54 y (44–64) Female: 33%	Patients in therapeutic doses excluded. Prophylaxis 8–10 days post-surgery	Technothrombin At inclusion	Symptomatic VTE confirmed by imaging or autopsy report VTE incidence: 17% in 2 y FU (n = 24) PE (46%), DVT (54%)	TG ↔ in VTE vs. non-VTE: Peak for VTE group: 480 nM (263–670) Peak for whole group: 432.8 nM (IQR: 192.8–597.0), p NR HR for VTE development per 50 nM increase in peak: Univariate HR  = 1.04 (CI: 0.96–1.16), p = 0.311. Multivariate HR = 1.03 (0.95–1.14), p = 0.451, both
Undas et al 2015[43] Good	Prospective observational	Newly diagnosed multiple myeloma patients undergoing induction therapy (n = 48) Age: 62 y (56–71) Female: 63%	No anticoagulant treatment allowed prior to inclusion. Aspirin or enoxaparin as thromboprophylaxis during chemotherapy (distribution NR)	Thrombinoscope At inclusion	Symptomatic VTE confirmed with imaging during the FU period VTE-incidence: 21% (n = 10) in 3 mo FU PE (10%), DVT (90%)	TG ↑ in VTE[c] vs. non-VTE: Peak: 503.5 (418–550) vs. 344.8 (269–411), p <  0.001
Yerrabothala et al 2021[46] Fair	Prospective observational	Glioblastoma patients undergoing surgery (n = 20) Mean age: 64 y, range 37–83 Female: 35%	No anticoagulant treatment allowed at inclusion	Thrombinoscope At inclusion	Screening of medical records for incident VTE at FU VTE-incidence: 10% (n = 2) in 6 mo FU Type NR	TG ↑ in VTE vs. non-VTE: Peak: Mean 364 nM vs. 290 nM, p = 0.04

Abbreviations: AUC, area under the curve; CI, 95% confidence interval; CT, computed tomography; DVT, deep venous thrombosis; ELISA, enzyme-linked immunosorbent assay; ETP, endogenous thrombin potential; FU, follow-up; HR, hazard ratio; IQR, interquartile range; LMWH, low molecular weight heparin; MRI, magnetic resonance imaging; NR, not reported; NS, nonsignificant; OR, odds ratio; PE, pulmonary embolisms; ROC, receiver operating characteristic; RR, relative risk; SD, standard deviation; SE, standard error; TF, tissue factor; TG, thrombin generation; ttPeak, time to peak; VTE, venous thromboembolic event(s).

Note: Age and findings reported as median (IQR) unless stated otherwise.

Note: ↔ no change, ↑ increased, (↑) discreetly increased.

^a Study population part of the same cohort (the Vienna CATS-study).

^c Two patients who experienced an ischemic stroke have been included in this group, alongside the 10 patients with VTE.

Fig. 2 Flow chart of the inclusion and exclusion process. TG, thrombin generation; VTE, venous thromboembolic events. Figure created with BioRender.

Study Characteristics

Most studies were performed on patients with solid cancers (n = 17).[13] [15] [25] [26] [27] [29] [30] [31] [32] [34] [35] [36] [37] [38] [40] [42] [44] Studies on purely hematological cancer patients were few (n = 3).[39] [41] [43] The remaining four studies had mixed study groups.[19] [21] [28] [33] All the included patients were adults. Mean follow-up times ranged from 7 days[25] [37] to 2.5 years,[41] and most studies were prospective observational.

Two studies recorded recurrent VTE; in Sallah et al, VTE was an inclusion criterion and thus all recorded VTE events were recurrent.[30] Tsubata et al screened for VTE at inclusion and allowed patients with a VTE to participate in the study. They recorded both incidence and recurrence at 2 years follow-up.[32] The remaining studies excluded patients with prior recorded VTE events.

Seven studies were performed on cancer patients undergoing surgery[25] [26] [27] [36] [37] [38] [44] investigating postoperative VTE, and five studies were performed in cohorts where >30% of the population underwent surgery.[19] [21] [30] [34] [35] In nine studies, all the patients received chemotherapy,[13] [15] [29] [34] [39] [40] [41] [42] [43] and in five, studies some patients received chemotherapy.[19] [21] [30] [35] [38] Thromboprophylaxis was allowed in 13 studies,[15] [19] [21] [25] [26] [27] [29] [31] [36] [38] [39] [40] [44] and therapeutic doses were allowed in four studies.[30] [32] [37] [41]

In total, 10 out of 11 studies investigating F1.2 used the same enzyme immunoassay. Most studies investigating TAT used the same enzyme immunoassay.[25] [26] [27] [33] [34] Two studies used an automated immunochemistry analysis[35] or a chemiluminescent immunoassay,[37] and the remaining two studies did not specify which assay they used.[30] [36] Studies investigating thrombin generation utilized either the Calibrated Automated Thrombogram (Thrombinoscope)[13] [38] [39] [41] [42] [43] [44] [45] or the similar Thrombin Generation Assay (Technothrombin).[21] [28] [29] [31]

Prothrombin Factor 1.2

Studies that were part of the Vienna CATS cohort generally found significantly increased F1.2 levels in patients who later developed VTE[19] [28] [29] [31] ([Table 2]). Ay et al found a median F1.2 in the patients who later developed VTE of 310 versus 249 pmol/L, yielding a hazard ratio (HR) of 2.0[19] (confidence interval [CI]: 1.2–3.6). These results were supported by Posch et al,[28] Reitter et al,[29] and Thaler et al,[31] who found HRs ranging from 1.48 to 2.11 for the development of VTE with F1.2 above cut-off (p-values from <0.001–0.069).

The larger study by Iversen and Thorlacius-Ussing[27] (n = 137) did not find significantly elevated preoperative F1.2 levels in patients who later developed a postoperative VTE, but the difference between the groups became statistically significant on postoperative day 1 (p < 0.002) and day 2 (p < 0.01). In the smaller, initial study (n = 93) by the same authors no difference was demonstrated.[26] Tsubata et al[32] found significantly increased F1.2 levels in their VTE cohort, yielding an odds ratio (OR) of 1.12 (CI: 1.073–1.169). However, this cohort included both incident and recurrent VTE. The remaining four studies[15] [25] [30] [33] found only weak and nonsignificant associations between F1.2 and the development of CAT.

Thrombin-Antithrombin Complex

Three of the included nine studies found higher levels of TAT in patients who later developed VTE[25] [35] [36] than in those who did not develop VTE ([Table 3]). Falanga et al[25] found 7.5 times increased relative risk (RR) of VTE in patients with a TAT > 3.5 ng/mL (p < 0.01). Similar results were found by Qi et al,[36] where patients with increased TAT had an RR of 6.12 for the development of VTE (p-value not reported). Liang et al[35] did not calculate a risk estimate but found significantly elevated TAT levels in their VTE cohort versus the patients who did not develop VTE. Sallah et al[30] investigated recurrent VTE and found an OR of 1.57 for VTE in their recurrence cohort (p = 0.01). Similar to their findings on F1.2, Iversen et al[26] [27] did not find an association between preoperative TAT levels and postoperative VTE but did demonstrate an increased TAT on postoperative day 1 in VTE patients compared to those that did not develop VTE ([Table 3]). Takata et al[37] had similar findings in their study on the incidence of portal vein thrombosis following hepatectomy, finding a nonsignificant postsurgery spike in TAT ([Table 3]). The resulting OR for developing VTE with an increased TAT was 1.20 (cut-off NR, CI: 1.01–1.42), p < 0.05). Finally, Kirwan et al and van Doormaal et al also found statistically nonsignificant increased TAT levels in patients who later developed VTE.[33] [34]

Ex Vivo Thrombin Generation

Four studies on ex vivo thrombin generation originated from the Vienna CATS cohort[21] [28] [29] [31] ([Table 4]). Of these, the largest study by Ay et al[21] (n = 1,033) found the strongest association between thrombin generation and development of VTE. They demonstrated overall increased thrombin generation in the VTE group, demonstrated by decreased time to peak thrombin generation and lagtime as well as an increased velocity index and peak thrombin in the patients who developed VTE versus those who did not. This translated to an HR of 1.9 for VTE for peak thrombin values greater than the 75th percentile, adjusted for age, sex, surgery, chemo- or radiation therapy, tumor type, and stage (p = 0.008). These results were in line with Posch et al,[28] who found a 46% increased risk of VTE for 1 SD increase in measured peak thrombin (n = 804, p = 0.01). The two smaller CATS studies[29] [31] (n = 112 and n = 141) found no increased VTE risk associated with higher peak thrombin measurements. Despite a small population size (n = 33), Leiba et al[41] demonstrated highly significant increased ETP of 2,896 versus 2,028 nM x min (p < 0.001) and peak thrombin of 620 versus 400 nM (p < 0.001) in multiple myeloma patients developing VTE. A similar result was found by Undas et al,[43] who also investigated patients with multiple myeloma starting chemotherapy (peak thrombin = 503.5 vs. 344.8, p < 0.001). The incidence of VTE was 21 to 24% in the two studies. Syrigos et al[42] found increased OR for the development of VTE in lung cancer patients with increased thrombin generation, but only the mean rate index-based OR = 1.016 reached statistical significance.

In the studies by Abu Saadeh et al[38] and Gezelius et al,[40] the cohort developing VTE had nonsignificantly increased thrombin generation. The study by Chalayer et al[39] yielded conflicting, nonsignificant results with an increased peak thrombin in the VTE group (186 vs. 149 nmol/L) but longer time to peak thrombin generation (10.8 vs. 9 minutes) in the same cohort. A similar result was found in the Schorling et al[13] study, where the VTE cohort had lower peak thrombin values (290 vs. 380 nM, p = 0.005) but higher ETP (values NR, p = 0.036) than the cohort without VTE.

Addition of Thrombin Generation Markers to Risk Assessment Scores

Pépin et al[15] investigated the addition of F1.2 measurements to the Khorana and Vienna CATS score. Both scores were improved when F1.2 was added, with an OR of 5.7 for the Khorana score and an OR 3.2 for the Vienna CATS score. When the authors added F1.2 and ADAMTS-13 to the CATS score, area under the receiver operator characteristics (AUROC) curve improved from 0.56 to 0.71. Ay et al[19] found an HR of 2.0 (p = 0.02) for the development of VTE in patients with F1.2 above the 75th percentile of the study population but did not apply this finding to an established risk assessment score. Tsubata et al[32] developed a score including F1.2 and found this to be a better predictor of VTE than fibrin D-dimer, with an AUROC curve of 0.75 (p < 0.001). Thaler et al[31] also developed a score, but neither peak thrombin generation nor F1.2 was a strong enough predictor for VTE to be included. None of the included studies investigated the addition of TAT to a risk assessment score.

The addition of mean rate index and procoagulant phospholipid-dependent clotting time (Procoag-PPL) to the COMPASS-CAT score improved positive predictive value from 13 to 70%, sensitivity from 83 to 88%, and specificity from 51 to 70% in one study.[42] The negative predictive value was unchanged at 97%. By itself, the mean rate index had an OR of 1.02 (p = 0.02).[42] Schorling et al[13] examined the Khorana, Vienna CATS, PROTECHT, ONKOTEV score, and Catscore, but found no improvement by the addition of ex vivo thrombin generation parameters in their cohort.

Discussion

Overall, the results from the present review presents evidence for the association between elevated markers of thrombin generation and VTE in patients with cancer. We chose to investigate F1.2, TAT, and ex vivo thrombin generation, as they all are well-established markers of thrombin generation. Of these, F1.2 yielded the most consistent and convincing result. Generally, studies with large cohorts demonstrated significantly elevated F1.2 levels in patients who later developed VTE. The smaller studies with nonsignificant findings pointed in the same direction but were probably underpowered to show significance. This gives merit to the idea of including F1.2 in risk assessment scores to improve sensitivity and/or specificity. Indeed, two of the included studies demonstrated an improvement of the Khorana and Vienna CATS score when F1.2 was added.[15] [32] Iversen et al. did not demonstrate a difference in their initial study obtaining blood samples prior to surgery[26] but found significant differences in their later, larger study[27] on the same population also analyzing F1.2 levels on postoperative days 1 and 2. Whether measurements taken after surgery may be better for risk assessment is an interesting topic for future studies. Overall, F1.2 seems to be able to improve prediction of VTE in larger cancer cohorts.

TAT was investigated in slightly fewer studies than F1.2. This, combined with greater assay heterogeneity, makes results more explorative in nature. Overall, TAT seemed to follow F1.2 and increase in patients at greater risk of VTE. Iversen and Thorlacius-Ussing[27] and Takata et al[37] reported greater difference in TAT levels between the patients who developed VTE versus the patients who did not, on postoperative day 1 and 2 compared to the preoperative sample. Like with F1.2, this seems justify a call for future studies examining the importance of sample timing on VTE association in surgical cancer patients. Overall, TAT seems to be elevated in patients developing VTE, though not as pronounced as F1.2.

Results on increased ex vivo thrombin generation and development of VTE in the included studies were heterogenous. However, most included studies showed increased ex vivo thrombin generation in cancer patients developing VTE. Thus, the overall results were in line with those from F1.2 and TAT. Interestingly, two of the studies that found strong associations between ex vivo thrombin generation and the development of VTE were in cohorts of patients with multiple myeloma.[41] [43] This may be due to a high VTE incidence in these two studies (21–24%).

Two studies found conflicting results, reporting lower ex vivo thrombin generation in patients who later developed VTE.[13] [39] In the study by Chalayer et al,[39] use of thromboprophylaxis was not found to be associated with any thrombin generation parameter. The study by Schorling et al[13] did not allow use of anticoagulants. The authors conclude that ex vivo thrombin generation may be less useful as a biomarker than F1.2 and TAT as the analysis only displays the thrombin potential and not the actual in vivo turnover. Furthermore, measurement of ex vivo thrombin generation is arguably less standardized than the F1.2 and TAT assays. Overall, then, ex vivo thrombin generation seems to be a weaker biomarker for VTE risk in cancer patients than F1.2 and TAT.

As evidenced in several studies, the risk of VTE is highly variable between different cancer types and the prevalence of additional VTE risk factors such as age, sex, comorbidities and treatment may also vary.[13] [15] [19] Thus, it is probably not feasible to develop a one-size-fits-all risk assessment score, even with a score which takes the type of cancer into account in the points assigned. Rather, scoring systems should be developed for specific cancer types or groups.[43] If risk assessment scores are utilized in large diverse patient groups, it might be prudent to aim to predict patients in low risk of VTE that thus might avoid thromboprophylaxis, instead of aiming to identify high-risk patients.[31] [34]

The included studies ranged quite widely in quality and comparability, which of course impacts the validity of the overall findings of this review. However, despite this, there is relatively high consensus in the findings which support the overall conclusions drawn.

Some variability in sample processing and analysis between studies was noted. For instance, Leiba et al[41] used a shorter centrifugation time, and Iversen et al[26] [27] and Kirwan et al[34] utilized cooled centrifugation, which may impact results. However, the number of commercially available assays are limited, and the analysis principles are largely similar within the three analyses. Thus, there was an overall high degree of method comparability within F1.2, TAT, and ex vivo thrombin generation studies which made interstudy comparison possible and credible.

As evidenced in the included studies, the definition of VTE varied greatly.[31] [37] [42] Two studies included recurrent VTE.[30] [32] Some studies included only symptomatic or accidentally discovered VTE and did not screen the participants systematically. While this may be due to logistical or financial considerations rather than a lack of interest in asymptomatic VTE, it makes results in unscreened cohorts harder to interpret. However, the clinical significance of incidental VTE is not fully understood. In the present review, the authors chose to exclude studies accepting device-related and isolated superficial VTE as events. Arterial events were also excluded, as the mechanisms behind arterial and venous thrombosis in cancer may differ.[4] This choice was made in attempt to create a more homogenous population but of course limits the conclusions in this review to venous thromboses.

The use of thromboprophylaxis was generally not well reported in the included studies, and few studies reported on the proportion of patients receiving thromboprophylaxis in the VTE versus non-VTE group or the mean duration of thromboprophylaxis. This made it difficult to consider the influence of thromboprophylaxis on the reported associations, especially for ex vivo thrombin generation. However, the included studies that allowed anticoagulant treatment or prophylaxis generally also found a signification association between markers of thrombin generation and development of VTE, thus strengthening the evidence for the association.

Some of the included studies enrolled patients undergoing chemotherapy and/or surgery. While this reflects a real-world cancer population, both chemotherapy and surgery are known to affect coagulation and therefore may complicate result interpretation. A variable proportion of the patients received anticoagulant treatment or prophylaxis in the observation period.

The selected biomarkers are all reasonably replicable and can be used on stored samples. This makes them more accessible for research as well as future routine implementation. F1.2 and TAT may be easier to automate and incorporate in a routine laboratory setup than ex vivo thrombin generation as they can be measured with widely available enzyme-linked immunosorbent assay assays and results are reported as concentrations which simplifies interpretation.[15]

Conclusion

The present review finds consistent evidence of increased markers of thrombin generation in cancer associated VTE. F1.2 appears to be the most promising predictor of the three selected markers and demonstrably increases the precision of established risk assessment scores.

Conflict of Interest

T.G. declares no conflict of interest. A.M.H. has no conflicts of interest regarding the present paper but has received an unrestricted research grant from CSL Behring. T.D.C. has been on the speaker bureaus for AstraZeneca, Boehringer-Ingelheim, Pfizer, Roche Diagnostics, Takeda, Merck Sharp & Dohme (MSD) and Bristol-Myers Squibb and has been in an Advisory Board for Bayer, Merck Sharp & Dohme (MSD), AstraZeneca and Sanofi. J.B.L. has received speaker's honoraria from Bristol-Myers Squibb and travel support from Bayer.

• References

• 1 Van Haren RM, Valle EJ, Thorson CM. et al. Long-term coagulation changes after resection of thoracoabdominal malignancies. J Am Coll Surg 2014; 218 (04) 846-854
• 2 Fernandes CJ, Morinaga LTK, Alves Jr JL. et al. Cancer-associated thrombosis: the when, how and why. Eur Respir Rev 2019; 28 (151) 180119
• 3 Khorana AA, Francis CW, Culakova E, Kuderer NM, Lyman GH. Thromboembolism is a leading cause of death in cancer patients receiving outpatient chemotherapy. J Thromb Haemost 2007; 5 (03) 632-634
• 4 Costamagna G, Navi BB, Beyeler M, Hottinger AF, Alberio L, Michel P. Ischemic stroke in cancer: mechanisms, biomarkers, and implications for treatment. Semin Thromb Hemost 2023; 50 (03) 342-359
• 5 Khorana AA, Francis CW, Kuderer NM. et al. Dalteparin thromboprophylaxis in cancer patients at high risk for venous thromboembolism: A randomized trial. Thromb Res 2017; 151: 89-95
• 6 Timp JF, Braekkan SK, Versteeg HH, Cannegieter SC. Epidemiology of cancer-associated venous thrombosis. Blood 2013; 122 (10) 1712-1723
• 7 Khorana AA, Kuderer NM, Culakova E, Lyman GH, Francis CW. Development and validation of a predictive model for chemotherapy-associated thrombosis. Blood 2008; 111 (10) 4902-4907
• 8 Ay C, Dunkler D, Marosi C. et al. Prediction of venous thromboembolism in cancer patients. Blood 2010; 116 (24) 5377-5382
• 9 Verso M, Agnelli G, Barni S, Gasparini G, LaBianca R. A modified Khorana risk assessment score for venous thromboembolism in cancer patients receiving chemotherapy: the PROTECHT score. Intern Emerg Med 2012; 7 (03) 291-292
• 10 Cella CA, Di Minno G, Carlomagno C. et al. Preventing venous thromboembolism in ambulatory cancer patients: the ONKOTEV study. Oncologist 2017; 22 (05) 601-608
• 11 Pabinger I, van Es N, Heinze G. et al. A clinical prediction model for cancer-associated venous thromboembolism: a development and validation study in two independent prospective cohorts. Lancet Haematol 2018; 5 (07) e289-e298
• 12 Gerotziafas GT, Taher A, Abdel-Razeq H. et al; COMPASS–CAT Working Group. A predictive score for thrombosis associated with breast, colorectal, lung, or ovarian cancer: the prospective COMPASS-cancer-associated thrombosis study. Oncologist 2017; 22 (10) 1222-1231
• 13 Schorling RM, Pfrepper C, Golombek T. et al. Evaluation of biomarkers for the prediction of venous thromboembolism in ambulatory cancer patients. Oncol Res Treat 2020; 43 (09) 414-427
• 14 Guman NAM, van Geffen RJ, Mulder FI. et al. Evaluation of the Khorana, PROTECHT, and 5-SNP scores for prediction of venous thromboembolism in patients with cancer. J Thromb Haemost 2021; 19 (12) 2974-2983
• 15 Pépin M, Kleinjan A, Hajage D. et al. ADAMTS-13 and von Willebrand factor predict venous thromboembolism in patients with cancer. J Thromb Haemost 2016; 14 (02) 306-315
• 16 Larsen JB, Hvas AM. Thrombin: a pivotal player in hemostasis and beyond. Semin Thromb Hemost 2021; 47 (07) 759-774
• 17 Zwicker JI, Furie BC, Furie B. Cancer-associated thrombosis. Crit Rev Oncol Hematol 2007; 62 (02) 126-136
• 18 Reddel CJ, Tan CW, Chen VM. Thrombin generation and cancer: contributors and consequences. Cancers (Basel) 2019; 11 (01) 100
• 19 Ay C, Vormittag R, Dunkler D. et al. D-dimer and prothrombin fragment 1 + 2 predict venous thromboembolism in patients with cancer: results from the Vienna Cancer and Thrombosis Study. J Clin Oncol 2009; 27 (25) 4124-4129
• 20 Chen Q, Shou W, Wu W, Wang G, Cui W. Performance evaluation of thrombomodulin, thrombin-antithrombin complex, plasmin-α2-antiplasmin complex, and t-PA: PAI-1 complex. J Clin Lab Anal 2019; 33 (06) e22913
• 21 Ay C, Dunkler D, Simanek R. et al. Prediction of venous thromboembolism in patients with cancer by measuring thrombin generation: results from the Vienna Cancer and Thrombosis Study. J Clin Oncol 2011; 29 (15) 2099-2103
• 22 Tripodi A. Thrombin generation assay and its application in the clinical laboratory. Clin Chem 2016; 62 (05) 699-707
• 23 Page MJ, McKenzie JE, Bossuyt PM. et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 2021; 372 (71) n71
• 24 National Heart L, and Blood Institute. Study quality assessment tools. Accessed 14.04. at: https://www.nhlbi.nih.gov/health-topics/study-quality-assessment-tools
• 25 Falanga A, Ofosu FA, Cortelazzo S. et al. Preliminary study to identify cancer patients at high risk of venous thrombosis following major surgery. Br J Haematol 1993; 85 (04) 745-750
• 26 Iversen LH, Okholm M, Thorlacius-Ussing O. Pre- and postoperative state of coagulation and fibrinolysis in plasma of patients with benign and malignant colorectal disease—a preliminary study. Thromb Haemost 1996; 76 (04) 523-528
• 27 Iversen LH, Thorlacius-Ussing O. Relationship of coagulation test abnormalities to tumour burden and postoperative DVT in resected colorectal cancer. Thromb Haemost 2002; 87 (03) 402-408
• 28 Posch F, Thaler J, Zlabinger GJ. et al. Soluble vascular endothelial growth factor (sVEGF) and the risk of venous thromboembolism in patients with cancer: results from the Vienna Cancer and Thrombosis Study (CATS). Clin Cancer Res 2016; 22 (01) 200-206
• 29 Reitter EM, Kaider A, Ay C. et al. Longitudinal analysis of hemostasis biomarkers in cancer patients during antitumor treatment. J Thromb Haemost 2016; 14 (02) 294-305
• 30 Sallah S, Husain A, Sigounas V. et al. Plasma coagulation markers in patients with solid tumors and venous thromboembolic disease receiving oral anticoagulation therapy. Clin Cancer Res 2004; 10 (21) 7238-7243
• 31 Thaler J, Ay C, Kaider A. et al. Biomarkers predictive of venous thromboembolism in patients with newly diagnosed high-grade gliomas. Neuro-oncol 2014; 16 (12) 1645-1651
• 32 Tsubata Y, Hotta T, Hamai K. et al. A new risk-assessment tool for venous thromboembolism in advanced lung cancer: a prospective, observational study. J Hematol Oncol 2022; 15 (01) 40
• 33 van Doormaal F, Kleinjan A, Berckmans RJ. et al. Coagulation activation and microparticle-associated coagulant activity in cancer patients. An exploratory prospective study. Thromb Haemost 2012; 108 (01) 160-165
• 34 Kirwan CC, McDowell G, McCollum CN, Kumar S, Byrne GJ. Early changes in the haemostatic and procoagulant systems after chemotherapy for breast cancer. Br J Cancer 2008; 99 (07) 1000-1006
• 35 Liang H, Guo X, Li Y. Cervical cancer associated biomarkers of identify high risk of venous thrombosis. Clin Lab 2020; 66 (05)
• 36 Qi Y, Hu X, Chen J, Ying X, Shi Y. The risk factors of VTE and survival prognosis of patients with malignant cancer: implication for nursing and treatment. Clin Appl Thromb Hemost 2020; 26 (e-pub ahed of print)
• 37 Takata H, Hirakata A, Ueda J. et al. Prediction of portal vein thrombosis after hepatectomy for hepatocellular carcinoma. Langenbecks Arch Surg 2021; 406 (03) 781-789
• 38 Abu Saadeh F, Langhe R, Galvin DM. et al. Procoagulant activity in gynaecological cancer patients; the effect of surgery and chemotherapy. Thromb Res 2016; 139: 135-141
• 39 Chalayer E, Tardy-Poncet B, Karlin L. et al. Thrombin generation in newly diagnosed multiple myeloma during the first three cycles of treatment: an observational cohort study. Res Pract Thromb Haemost 2018; 3 (01) 89-98
• 40 Gezelius E, Flou Kristensen A, Bendahl PO. et al. Coagulation biomarkers and prediction of venous thromboembolism and survival in small cell lung cancer: a sub-study of RASTEN—-a randomized trial with low molecular weight heparin. PLoS One 2018; 13 (11) e0207387
• 41 Leiba M, Malkiel S, Budnik I. et al. Thrombin generation as a predictor of thromboembolic events in multiple myeloma patients. Blood Cells Mol Dis 2017; 65: 1-7
• 42 Syrigos K, Grapsa D, Sangare R. et al. Prospective assessment of clinical risk factors and biomarkers of hypercoagulability for the identification of patients with lung adenocarcinoma at risk for cancer-associated thrombosis: the observational ROADMAP-CAT study. Oncologist 2018; 23 (11) 1372-1381
• 43 Undas A, Zubkiewicz-Usnarska L, Helbig G. et al. Induction therapy alters plasma fibrin clot properties in multiple myeloma patients: association with thromboembolic complications. Blood Coagul Fibrinolysis 2015; 26 (06) 621-627
• 44 Yerrabothala S, Gourley BL, Ford JC. et al. Systemic coagulation is activated in patients with meningioma and glioblastoma. J Neurooncol 2021; 155 (02) 173-180
• 45 Gezelius E, Belting M. Biomarkers of venous thromboembolism in cancer: a silent echo from local events?. Biomarkers Med 2019; 13 (07) 507-509
• 46 Mrozinska S, Cieslik J, Broniatowska E, Malinowski KP, Undas A. Prothrombotic fibrin clot properties associated with increased endogenous thrombin potential and soluble P-selectin predict occult cancer after unprovoked venous thromboembolism. J Thromb Haemost 2019; 17 (11) 1912-1922

Publication History

Article published online:
09 October 2023