1. Introduction
Cancer patients often present with a prothrombotic state due to the abnormalities in each component of Virchow’s triad, thus contributing to thrombosis. Researchers estimated that VTE would occur in 4–20% of cancer patients at some stage, with the highest risk immediately following cancer diagnosis [ 1 ]. In the last period, the VTE incidence in oncologic patients has increased in the context of the higher performance of imaging techniques and the development of new cancer treatments that improved survival [ 2 ]. After cancer diagnosis, the 12-month cumulative VTE incidence was 3%, a percentage nine times higher when compared to the general population [ 2 ].
However, despite improved cancer treatment, VTE in cancer patients is strongly associated with a poor prognosis. The cumulative mortality in VTE cancer patients was 27.7% after one month, 48.7% after three months, 68.2% at one year, and 84.1% after five years, which is much higher than the cumulative mortality in cancer patients without VTE (7.5%, 17%, 38.5%, and 84.1%, respectively) [ 3 ]. Pulmonary embolism (PE) was associated with a poorer prognosis than venous thrombosis [ 3 ]. The one-year mortality of the PE cancer patients was 73% in Sørensen et al.’s study, as compared to 39.3% in the non-cancer cohort [ 3 ].
Khorana et al. reported that 17.1% of the patients recently diagnosed with cancer and with VTE events would develop recurrent episodes of VTE during a nine-month follow-up period [ 4 ]. The total costs related to the healthcare of the patients with VTE recurrence were very high, suggesting the necessity of reducing VTE risk in cancer patients [ 4 ].
Thromboprophylaxis in hospitalized cancer patients and perioperative settings is widely accepted in clinical practice and supported by the guidelines. However, most cancer patients would develop VTE in the outpatient setting. Primary thromboprophylaxis is not routinely recommended, except for high-risk cancer patients. Selecting an ambulatory cancer patient who would benefit from thromboprophylaxis is still challenging because of the specific bleeding risk. The Khrorana score is mainly the recommended tool in this setting, but many limits of this old score have been reported. Novel approaches have been proposed. Clinical features, routine hematologic and coagulation lab testing, new biomarkers, and genetic data, separately or grouped, were introduced in the novel risk scores, nomograms, or machine learning algorithms to accurately assess the VTE risk in ambulatory cancer patients in general and specific tumors. This narrative review aims to present the current relevant knowledge in this setting starting from the guideline recommendations and continuing with the specific risk assessment methods to help clinicians in their decision regarding primary thromboprophylaxis in ambulatory cancer patients ( Figure 1 ). The future directions provided by the recent research papers are also presented.
2. Guideline Recommendations
The European Society for Medical Oncology (ESMO) 2010 practical guidelines proposed the Khrorana model to identify ambulatory cancer patients who are clinically at high risk for VTE [ 5 ]. The new 2023 ESMO guideline suggested the use of the same Khrorana model, but also the Vienna-CATS and COMPASS-CAT methods [ 6 ].
The same guideline strongly recommended ultrasound diagnosis and computer tomography (CT) pulmonary angiogram if deep venous thrombosis (DVT) or pulmonary embolism (PE), respectively, were suspected in cancer patients [ 6 ]. The D-dimer levels and the clinical prediction rules have not been recommended in this setting [ 6 ].
The new 2023 American Society of Clinical Oncology (ASCO) guideline considers only the Khrorana score in stratifying the risk in this setting [ 7 ]. The American Society of Hematology (ASH) 2021 guidelines provided strong and conditional recommendations for not using thromboprophylaxis on low and intermediate-risk ambulatory oncologic patients receiving cancer chemotherapy, respectively [ 8 ]. A validated risk assessment tool (i.e., Khorana score) together with clinical judgment and experience were recommended for patient classification [ 8 ]. The hereditary thrombophilia tests were suggested by the 2023 update in ambulatory cancer patients receiving systemic therapy with VTE family history determined to be at low or intermediate risk for VTE [ 9 ].
The European Society of Cardiology (ESC) 2022 cardio-oncology guideline recommended the baseline clinical and biomarkers assessment of the patient diagnosed with cancer [ 10 ]. Multidisciplinary monitoring during the specific treatments was also proposed [ 10 , 11 ]. The ESC 2022 cardio-oncology guideline emphasized the high incidence of VTE among cancer patients and recommended imaging screening in patients with clinically suspected VTE [ 10 ]. Lower-extremity venous ultrasonography is the method that has to be used in DVT diagnosis, as well as contrast-enhanced CT [ 10 ]. After detailing the patient-, cancer-, and treatment-related risk factors, the guideline proposed the TBIP method in the anticoagulation decision. It represents an acronym for thromboembolic risk, bleeding risk, drug–drug interactions, and patient preferences [ 10 ].
The ESC guideline recommended that the VTE risk assessment in ambulatory patients be individually determined and only found the Khorana and COMPASS-CAT scores useful in this setting [ 10 ]. Khrorana risk assessment was also recommended by the American College of Cardiology in ambulatory oncologic patients [ 12 ].
Table 1 summarizes the relevant guideline recommendations on VTE risk assessment in ambulatory oncologic patients.
3. VTE Screening in Ambulatory High-Risk Oncologic Patients
Cancer and thrombosis are strongly related. VTE can be the first clinical sign of undiagnosed cancer, especially when the event is unprovoked [ 13 ], while cancer represents a risk factor in VTE occurrence. In this last setting, guidelines issued recommendations for hospitalized and surgical cancer patients and high-risk outpatients. Gainsbury et al. found a 10.1% prevalence of preoperative deep venous thrombosis (DVT) in asymptomatic patients undergoing major oncologic surgery and suggested the preoperative screening with lower extremity venous duplex ultrasound (US) in this setting [ 14 ]. Increasing age, recent diagnosis of sepsis, and a history of prior VTE were significantly associated with preoperative DVT [ 14 ].
Detecting VTE high-risk outpatients with cancer is still challenging. VTE screening may be an answer in this setting. In total, 6.6% of venous thrombosis was found by Heidrich et al. in all tumor patients [ 15 ]. The same authors reported a much higher incidence of 33% when using an imaging prospective approach [ 15 ]. Loftus et al. researched the role of venous US screening in incidentally detecting VTE in high-risk patients with cancer in a multicenter trial. The studied 117 patients were asymptomatic, had a Khorana score ≥ 3, and were starting new systemic chemotherapy [ 16 ]. The lower-limb venous US and a contrast-enhanced CT baseline screening discovered 9% incidental VTE (6% DVT, 1% pulmonary embolism, 1% DVT and pulmonary embolism) [ 16 ]. The patients were screened further every four weeks for a 12-week period with venous US and at 12 weeks with contrast-enhanced CT [ 16 ]. Researchers proposed the lower-limb venous US screening in addition to the oncologic surveillance CT in high-risk ambulatory cancer patients setting with a Khorana score ≥ 3 [ 16 ].
This approach could help in early VTE detection in latent stages, preventing VTE progression and thus decreasing morbidity and costs [ 16 ]. Kourlaba et al. also reported US screening of high-risk cancer patients as a cost-effective strategy compared to clinical surveillance, even when all patients with a positive first US underwent a second US [ 17 ]. Kunapareddy et al. proposed an electronic alert to identify high-risk patients and suggest US screening for early detection [ 18 ]. Holmes et al. reported the success of a multidisciplinary program related to Venous Thromboembolism Prevention in the Ambulatory Cancer Clinic (VTEPACC) [ 19 ]. The high-risk patients identified by Khorana and Protecht scores (≥3 points) were offered a hematology consultation to consider VTE prophylaxis, further referring the results of the consultation to the oncologist [ 19 ].
VTE risk was predicted by baseline D-dimer levels [ 20 , 21 ]. Niimi et al. recently reported the optimal D-dimer cut-off value of 4.0 μg/mL for predicting DVT in patients with malignancy [ 22 ]. Its association with risk assessment scores performed better in VTE prediction [ 21 , 22 ]. D-dimer was reported in another study as part of the thromboembolism risk assessment when added to fibrinogen level [ 23 ]. Oi et al. found that high D-dimer levels at VTE diagnosis were associated with an increased risk for short-term and long-term mortality and with long-term recurrent VTE, especially in patients with active cancer [ 24 ]. During a median follow-up of 30 months, D-dimer positively correlated with the reoccurrence of VTE ( p = 0.0299) and mortality in cancer patients with VTE ( p < 0.0001) and without VTE ( p = 0.0008) [ 25 ]. D-dimer level positively correlated in Koch et al.’s study with VTE reoccurrence and mortality during a 30-month period [ 25 ]. The relationship with mortality was reported both in cancer patients who presented VTE and in cancer patients without VTE [ 25 ].
Another VTE risk factor is the soluble P-selectin (sP-selectin). A cut-off level of 53.1 ng/mL could predict VTE in cancer patients with no difference between tumor sites [ 26 ]. Zhang et al. recommended sP-selectin level for early identification of cancer-associated VTE and monitoring [ 27 ].
Khorana et al. recently studied the biomarkers distribution in patients with and without VTE diagnosed with cancer [ 28 ]. In the two groups, there were reported baseline lower levels of stromal cell-derived factor-1, thyroid-stimulating hormone, and monocyte chemotactic protein 4 and higher levels of growth hormone and interleukin-1 receptor type 1 [ 28 ]. ST2, IL-8, and C-reactive protein were significantly different between survivors and those who died [ 28 ].
Table 2 presents the relevant studies presenting modalities and importance of VTE screening among ambulatory cancer patients.
microRNAs (miRNAs) represent a promising class of biomarkers in VTE prediction in cancer, but until now, only a few small-sample-size studies, lacking external validation, have investigated their role in this setting [ 29 ]. The long non-coding RNAs (lncRNAs) may have a role as well in VTE pathogenesis [ 30 ]. Ten lncRNAs were implicated in VTE pathogenesis, but future research is needed in this setting [ 30 ].
Genetic assessment may help VTE risk stratify and prognostic in the cancer population. Thrombogenesis-related genetic polymorphisms are already studied in this setting and are integrated in specific risk scores alone, or together with clinical features. However, more prospective studies are required before clinical application.
4. VTE Risk Assessment Using Scores
Because the CAT risk factors are multifactorial, risk scores have been developed to find oncologic outpatients who need anticoagulation treatment. The Khorana score was the first proposed [ 31 ]. The type of cancer, some components of the complete blood count, and body mass index were assessed. A value ≥ 2 was retained by the guidelines as describing high-risk patients [ 6 , 7 , 32 ], although a value of more than 3 was initially proposed [ 31 ]. Khorana et al. validated the method in a cohort with 34.6% breast cancer patients and 18.9% lung cancer patients [ 31 ]. The rest of Khorana et al.’s cohort had colon, ovarian, gastric, and pancreatic cancers, lymphomas, and other tumor types [ 31 ]. Mulder et al.’s meta-analysis reported the Khorana score as a tool for selecting high-risk VTE in oncologic patients [ 33 ], and Akasaka-Kihara et al. validated it in the Japanese cancer population [ 34 ]. Ramos-Esquivel et al. recently found the Khorana score to perform an accurate categorization of VTE risk in ambulatory Hispanic patients who were newly diagnosed with solid tumors and were receiving systemic chemotherapy [ 35 ]. El-Sayed et al. reported a calculated VTE occurrence probability of 87.5% when using the Khorana score at cut-off levels ≥ 3 in patients with hematological malignancy [ 36 ]. However, many researchers found the universal use of the Khorana score in primary thromboprophylaxis risk assessment inappropriate. Ha et al. (2023) only partially validated the Khorana score in the Korean population [ 37 ]. Khorana could stratify the 6-month VTE risk only in selected cancer populations [ 33 , 38 , 39 , 40 ]. Verzeroli et al. (2023) recently found that the Khorana score was not able to discriminate between low and high VTE risk in newly diagnosticated metastatic cancer (non-small cell lung, gastric, colorectal, and breast cancers) for whom systemic chemotherapy was indicated [ 41 ]. This score was unable to stratify VTE risk in lung cancer [ 39 , 42 , 43 ], endometrial [ 40 ], MM [ 44 ], myeloid leukemia [ 45 ], hepatocellular carcinoma [ 46 ], uterine [ 47 ], or lymphoid malignancies [ 48 ]. Although this risk score was suboptimal in VTE risk prediction, other studies found it more useful in mortality prediction [ 41 , 43 ] or when a value ≥ 2 was tested [ 49 , 50 ] in the VTE setting.
Ay et al. [ 51 ] proposed a new VTE risk assessment method in patients by adding two biomarkers, D-dimer and sP-selectin, to the Khorana score. Higher D-dimer (cut-off 1.44 μg/mL) and sP-selectin (cut-off 53.1 ng/mL) levels were reported previously by the authors to be associated with VTE [ 20 , 26 ] in a Vienna Cancer and Thrombosis Study (Vienna CATS). In a multinational, prospective cohort study, the Vienna CATS method discriminated better than the Khorana score between low- and high-risk VTE patients [ 52 ]. The eligible patients were those with advanced cancer who underwent chemotherapy or had started chemotherapy in the previous three months [ 52 ]. In hematological malignancy, the calculated probability of VTE occurrence was the same when using Vienna CATS or Khorana score at cut-off values of ≥3 [ 36 ]. A value more than or equal to 3 of the Vienna CATS risk score was significantly associated with VTE complications in Japanese patients with advanced cancer who were receiving chemotherapy [ 53 ].
Verso et al. proposed another risk score, the PROTECHT score, that added gemcitabine and platinum-based chemotherapy to Khorana score variables [ 54 ]. Moik did not sustain the use of these variables in prediction models [ 55 ]. The gemcitabine therapy has not been associated with an increased VTE risk, while platinum-based treatment had only limited predictive value beyond tumor site category and D-dimer levels [ 55 ]. However, in van Es et al.’s study, the PROTECHT score performed better discrimination than the Khorana score in the VTE risk assessment [ 52 ]. Ramos-Esquivel et al. also found PROTECHT (cut-off 3) among the scores that could categorize the VTE risk in newly diagnosed solid tumors in the Hispanic population [ 56 ]. Other studies reported this score as suboptimal in VTE risk assessment [ 49 , 50 ], proposing a 2-value threshold to improve the results [ 49 ].
Another risk score proposed in this setting was CONKO [ 57 ]. From the Khorana score variables, BMI was replaced by WHO performance status [ 57 ]. In Qin et al.’s (2023) study, Khorana, Vienna CATS, PROTECHT, and CONKO risk scores moderately assessed the VTE risk in hospitalized metastatic colorectal cancer inpatients [ 58 ], but the prediction was enhanced when KRAS and BRAF mutations were added to the scores [ 58 ]. In ambulatory oncologic patients, the VTE risk stratification by CONKO was suboptimal [ 49 , 52 ], although in the Hispanic population the results were more encouraging [ 56 ]. HYPERSCAN study reported that CONKO scores significantly stratified patients for VTE risk, while the KRS and the PROTECHT failed in ambulatory lung cancer patients [ 59 ]. In this setting, future research is awaited. Yan et al.’s systematic review regarding VTE risk assessment models for use in ambulatory patients with lung cancer is expected to be published soon [ 60 ].
Papinger et al. proposed a modified Vienna CATS, the CATS/MICA score, that integrated the tumor site category and D-dimer level to predict the VTE risk in ambulatory patients with solid cancers [ 61 ]. The new Vienna CATS and the CONKO scores significantly stratified patients for VTE risk in lung cancer [ 59 ]. Verzeroli et al. reported that a modified Vienna CATS score > 60 points was an independent risk factor for mortality in outpatients with metastatic cancer during chemotherapy [ 41 ].
Gerotziafas et al. proposed the COMPASS-CAT model, another VTE risk assessment method in a breast, colorectal, lung, or ovarian cancer cohort [ 62 ]. COMPASS-CAT takes into account several aspects referring to the time the cancer was diagnosed, the stage of the disease, previous VTE occurrences, platelet count, the presence of central venous catheter and the cardiovascular risk factors, the specific anti-hormonal treatment or with anthracycline, and recent hospitalization in acute medical setting [ 62 ]. The model presented in Gerotziafas et al.’s study had a good sensitivity of 88% but a lower specificity of 52% [ 62 ]. There was a strong association between catheter-related thrombosis and high Khorana, PROTECHT, and COMPASS-CAT scores [ 63 ]. COMPASS-CAT better identified more patients in high-risk group in non-small cell lung cancer [ 42 ]. Rupa-Matysek et al. reported that the COMPASS-CAT model was the most accurate predictor of VTE in lung cancer patients, compared to Khorana, PROTECHT, and CONKO scores [ 64 ]. In a large retrospective external validation study, the COMPASS-CAT model had good negative predictive value, with moderate discrimination and poor calibration power [ 65 ]. Pestana et al. (2023) reported a high VTE risk when evaluated by the COMPASS-CAT model (score ≥ 7) in breast cancer patients [ 66 ] and proposed the combination of this model with IL-10 levels to improve the method [ 66 ].
Cella et al. proposed a new assessment tool, the ONKOTEV score, an easy-to-use and cost-effective model based on clinical information, avoiding highly selective biochemical parameters. The ONKOTEV score offers one point for a Khorana score > 2, previous venous thromboembolism, metastatic disease, and vascular/lymphatic macroscopic compression [ 67 ]. In their prospective study, the area under the curve of ONKOTEV over the Khorana score was reported at 3 months (71.9% vs. 57.9%, p = 0.001), 6 months (75.4% vs. 58.6%, p < 0.001), and 12 months (69.8% vs. 58.3%, p = 0.014) [ 67 ]. Cella et al. (2023) validated this model in the ONKOTEV-2, a multicenter prognostic study on ambulatory patients with solid tumors undergoing active treatments [ 68 ]. The most represented tumors were breast (18.1%), gastroesophageal adenocarcinoma (16.5%), colon (12.7%), lung (11.1%), rectum (10.8%), and pancreatic cancers (7.5%) [ 68 ]. Di Nisio et al. reported that the performance of the Khorana, PROTECHT, CONKO, and ONKOTEV scores improved at the threshold of 2 points, compared to 3 points [ 49 ]. The scores’ accuracy decreased over time, suggesting the need for periodic re-evaluation [ 49 ]. An ONKOTEV score ≥ 2 was associated with a higher VTE occurrence in patients with pancreatic cancer, including ambulatory ones [ 69 ]. ONKOTEV score performed better than PROTECHT, COMPASS-CAT, CONKO, Khorana, and the CATS/MICA score in VTE risk assessment in hospitalized medical patients with primary lung cancer [ 70 ]. An ONKOTEV score ≥ 2 was also a predictor of survival and thromboembolic events in cholangiocarcinoma [ 71 ].
Table 3 summarizes the relevant studies related to the scores used in VTE risk assessment in ambulatory cancer patients.
Zhang et al. studied the systemic immune-inflammation index (SII) and the prognosis nutritional index (PNI) in VTE prediction in gastrointestinal cancer patients [ 72 ]. The SII was an auxiliary diagnostic test for patients with venous thrombosis in general, with an AUC of 0.861 (95% CI: 0.820–0.902; p < 0.001), a sensitivity of 78.1%, and a specificity of 73.1% for an SII > 755.54 [ 73 ]. SII (cut-off 504.80) paired with PNI (cut-off 45.57) were part of the two nomograms proposed to predict VTE risk in gastrointestinal cancer [ 72 ]. Model A (age, tumor location, therapy, PNI, SII) and Model B (age, tumor location, therapy, PNI, SII and D dimer) presented an AUC of 0.806 (95% CI: 0.782–0.830) and 0.832 (95% CI: 0.810–0.855), respectively, as compared to Khorana score’s 0.592 (95% CI: 0.562–0.621) [ 72 ]. Zhang et al. also tested SII in lung cancer patients and developed a new nomogram model (the inflammatory marker, coagulation indicator, and tumor features) to perform an accurate prediction of VTE [ 74 ]. The SII cut-off was 851.51, and the new nomogram presented an AUC of 0.708, compared to the Khorana score’s 0.600 [ 74 ].
Li et al.’ new nomogram contains common data from the electronic health record, some demographic data (Asian/Pacific islander), the original Khorana score, but with cancer subtypes that were revised, and cancer and patient risk factors such as hormonal/target therapy, advanced cancer, previous VTE occurrence, recent hospitalization, and history of immobility [ 75 ].
Approximately 50% of cancer patients receiving modern systemic therapy were stratified into a high-risk group (a 6-month VTE risk of 8–10%) and the other half into a low-risk group (a 6-month VTE risk of 3%) [ 75 ]. This novel tool appeared generalizable in variate age, sex, and race/ethnicity subgroups but needs further validation [ 75 ].
In the ambulatory cancer population, there is no universal method for the VTE risk assessment. Thus, new specific risk scores have been developed. In newly diagnosed NSCLC outpatients who undergo chemotherapy, Thrombo-NSCLC (FVIII and sP-selectin values) predicted VTE significantly better than the Khorana score [ 76 ]. Gomez-Rosas et al. recently proposed a new risk tool, the Hypercan score, to stratify lung cancer patients for VTE and mortality risk [ 59 ]. This score contains information regarding ECOG performance and D dimer and stratifies the patients into low- and high-risk groups [ 59 ]. Using the Hypercan score, the cumulative incidence of VTE was 6% in the low- and 25% in the high-risk group [ 59 ]. Li et al. proposed and validated a new nomogram for VTE risk prediction in patients recently diagnosed with lung cancer [ 77 ]. Some clinical and therapeutic features and genetic parameters were incorporated into the new assessment system: overweight, adenocarcinoma, stage III-IV, central venous catheters, D-dimer levels ≥ 2.06 mg/L, prothrombin time ≥ 11.45 s, fibrinogen levels ≥ 3.33 g/L, triglyceride ≥ 1.37 mmol/L, ROS1 rearrangement, chemotherapy history and radiotherapy history [ 77 ]. In lymphoma patients, the ThroLy score was proposed. It was designed for both hospitalized and outpatient settings and included data referring to tumor spread (mediastinal involvement, extranodal localization), frailty (reduced mobility, BMI > 30 kg/m 2 ), the presence of previous arterial or venous thromboembolic events, anemia (hemoglobin level < 100 g/L), neutropenia [ 78 ]. ThroLy not only predicted VTE in Hodgkin lymphoma, but also survival [ 79 ]. It has been studied in diffuse large B-cell lymphoma settings as well [ 80 ]. A simplified model was proposed (high-risk with a score ≥ 3 and low-risk score < 3) [ 80 ]. Others did not find this score an accurate model for predicting VTE events in patients at higher risk of VTE [ 81 ]. An adapted TiC-Onco risk score to lymphoma settings was proposed by Bastos-Oreiro et al. [ 82 ] with promising results. The TiC-LYMPHO score incorporated the same genetic variables included in the TiC-ONCO score and some of the clinical variables associated with VTE in the studied population (the type of lymphoma according to the WHO classification, mediastinal involvement, Ann Arbor stage, bed rest for >3 days, and a family or personal history of VTE) [ 82 ]. López Sacerio et al. (2023) found five predictive factors—hypercholesterolemia, tumoral activity, use of thrombogenic drugs, diabetes mellitus, and immobilization—that were integrated into a predictor model of VTE in patients with hospitalized hematologic malignancies [ 83 ].
The IMPEDE VTE score (immunomodulatory agent; body mass index ≥ 25 kg/m 2 ; pelvic, hip or femur fracture; erythropoietin stimulating agent; dexamethasone/doxorubicin; Asian ethnicity/race; VTE history; tunneled line/central venous catheter; existing thromboprophylaxis) was developed and validated by Sanfilippo et al. (2019) as a VTE risk prediction score in multiple myeloma (MM) [ 84 ]. Recently, Sanfilippo et al. (2023) reported that adding D-dimer to the IMPEDE VTE score could improve VTE prediction among MM patients [ 85 ].
The PICOS score (primary tumors with high thrombogenicity, immobilization, chemotherapy, obesity, and steroid) was proposed by Wolpert et al. as a helpful tool for the identification of patients with brain metastasis at high risk for VTE [ 86 ].
Kubo et al. found that the D-dimer combined with the Glasgow prognostic score accurately predicted VTE in stage IIIC and IVA of ovarian cancer (AUC: 0.846; p < 0.001) [ 87 ]. D-dimer could significantly predict VTE in all gynecologic cancer patients. Optimal reported D-dimer cut-off values were 3.1, 3.2, and 3.9 μg/mL in cervical, endometrial, and ovarian cancer patients, respectively [ 87 ].
5. Genetic-Based Risk Assessment Scores
Polygenic risk scores do not change during the cancer course. Thus, they could be potential predictors of cancer-associated VTE independent of cancer type. Both Factor V Leiden and ABO gene mutations were reported as independent predictors of VTE occurrence in moderate to high-risk outpatients with cancer undergoing chemotherapy [ 88 ].
Lindström et al. (2019) reported the results of a large genome-wide association study (GWAS) and the first transcriptome-wide association study (TWAS) on VTE risk. GWAS meta-analysis identified 34 independent genetic signals for VTE risk with 14 newly reported associations [ 89 ]. TWAS identified five additional genetic loci not previously associated with VTE (SPSB1, ERAP1, RP11-747H7.3, RP4-737E23.2, and replicated SH2B3) [ 89 ]. The researchers demonstrated that a genetic risk score based on 37 VTE-susceptibility variants can identify a subset of the population at high risk for developing VTE [ 89 ].
De Haan et al. designed a genetic score to select VTE high-risk patients [ 90 ]. This method contained data on 31 single-nucleotide polymorphisms (SNPs) associated with an increased risk [ 90 ]. The 5-SNP score (rs8176719, rs6025, rs1799963, rs2066865, and rs2036914) was created by adding one-by-one the SNPs with the highest odds ratios of VTE and similarly discriminated high-risk patients as 31 SNPs regarding both incidental and recurrent VTE events [ 91 ]. VTE risk increased with the number of prothrombotic risk alleles, independent of the cancer diagnosis [ 91 ]. The presence of both prothrombotic risk alleles and cancer represented a highly elevated VTE risk factor [ 91 ].
Other studies did not find the 5-SNP score to be superior to the Khorana score [ 50 ]. Jakobsen et al. reported an elevated discriminative effect of the 5-SNP score on VTE risk by combining it with the mean platelet volume, but the results were not focused on the cancer population [ 92 ].
The 5, 37, 297, extended 297, and 100 SNPs prospectively identified those cancer patients at high risk for VTE development in a population-based study [ 93 ]. The tumor type has not influenced the result [ 93 ]. The 36,150 patients of the UK Biobank cohort diagnosed with hematological or solid cancer were studied from the genetic point of view regarding VTE risk. In the 12 months post-cancer diagnosis, the germline genetic markers accurately selected the patients with an increased double risk for VTE occurrence. Guman et al. demonstrated that the tumor type and polygenic scores’ performance were independent variables, as the latter remained consistent during the 12-month follow-up [ 93 ]. The VTE prediction was improved when the two variables were combined [ 93 ].
The TiC-Onco risk score integrated genetic (rs2232698, rs6025, rs5985, rs4524) and clinical risk factors. Muñoz et al. used it to identify patients with colorectal, esophagogastric, lung, or pancreatic cancer in the outpatient setting who are at high risk of VTE [ 94 ]. This method has to be followed at the moment cancer is suspected [ 94 ]. Its sensitivity was significantly higher than that of the Khorana, while the specificities of both scores were similar in the studied population [ 94 ]. TiC-LYMPHO, a modified TiC-Onco score, was proposed in lymphoma settings [ 82 ]. Neto et al. (2023) reported PROCR rs10747514 and RGS7 rs2502448 as valuable prognostic biomarkers regardless of VTE and significantly associated with the VTE risk in cervical cancer [ 95 ].
Recently, nine genetic variants (rs4524, rs6025, rs2232698, rs2227631, rs268, rs169713, rs11696364, rs5110, rs6003) were independently associated with VTE in outpatients with cancer [ 96 ]. Muñoz et al. developed and validated ONCOTHROMB, by combining this genetic profile with three clinical variables independently associated with VTE in outpatients with cancer. This score, with a higher AUC compared to the Khorana model (AUC, 0.781 vs. 0.580; p < 0.001) was recommended to be assessed at the moment cancer is suspected [ 96 ]. ONCOTHROMB presented a significantly higher sensitivity than Khorana (81.54% vs. 22.54%; p < 0.001), with a lower specificity (65.22% vs. 81.76%; p < 0.0001) [ 96 ].
6. Machine Learning Algorithms Tools
Artificial intelligence brings new methods to assess risk. Machine learning (ML) can develop many statistical algorithms that can learn from the pattern of the database. The highly flexible novel tools can discriminate better in a nonlinear setting [ 97 , 98 ]. The manual data analysis is eliminated when using ML algorithms, while a large volume of data can be reviewed, identifying more easily patterns and trends. The automatic and dynamic self-learning process leads to continuous improvement in decision making. ML may help save time and resources at the clinician level by reducing the data analysis time, optimizing medical decisions, and offering insights into other centers’ experiences and databases. Thus, the clinician may have additional time to spend with patients to understand better their needs and disease. Of course, the input data must be correct and large enough to obtain accurate results. The parameters and ML algorithms must be continuously developed and optimized. Otherwise, the probability of high errors is high. The users must define the acceptable margins of the statistical error because the ML algorithms approach represents a probabilistic process. The ethical challenges in collecting and handling data represent another ML algorithm limitation. Clinicians must understand the advantages and the disadvantages of using the ML approach as they are the interface to the patients [ 99 ]. Clinical judgment still has its role in our era. The medical doctor is the one able to discriminate between clinical changes by integrating medical and social data. Still, the clinician is the one who can interpret the results of ML algorithms in the patient’s context.
Machine learning algorithms can help assess the risk of venous thromboembolism (VTE) in ambulatory cancer patients. By accurately identifying high-risk VTE patients, healthcare professionals can provide them with the appropriate treatment. In Ferroni et al.’s (2017) study, a model based on multiple kernel learning (MKL) and random optimization (RO) was used to achieve this goal in chemotherapy-treated ambulatory cancer patients [ 100 ]. ML-RO-2 was the most accurate model compared to the Khorana score (positive likelihood ratio 1.68, negative likelihood ratio 0.24) [ 100 ]. ML-RO-2 presented an area under the precision–recall curve of 0.212, while the Khorana score’ area was only 0.096 [ 100 ]. The strongest association was related to the blood lipids and body mass index/performance status, while the weaker was related to the tumor site/stage and drugs [ 100 ]. The second best-performing model was ML-RO-3 [ 100 ]. A study conducted by Ferroni et al. in 2017 validated the ML-RO-2 and ML-RO-3 approaches as a low-cost method for assessing VTE risk in oncologic patients [ 101 ]. The f-measure, a metric used in ML, calculated as a harmonic mean of P (positive predictive value in ML) and R (sensitivity in ML), measured the effectiveness of a classifier algorithm. ML-RO-2 and ML-RO-3 presented higher f-measures (0.213 and, respectively, 0.211) than the Khorana score (f-measure: 0.100) [ 101 ]. The study involved 608 patients, with a mean age of 63 years, with 58% of them having relapsing/metastatic solid cancers [ 101 ]. The incidence of deep venous thrombosis was 5.3%, while pulmonary embolism was diagnosed in 1.8% of cases [ 101 ]. Xu et al. (2023) developed and validated a new clinical prediction model for VTE in gastric cancer patients based on support vector machine (SVM), one of the ML algorithms [ 97 ]. The model’s AUC, sensitivity, and specificity were 0.825, 0.710, and 0.802, respectively [ 97 ]. The top five predictors in the model were the clinical stage, the blood transfusion history, D-dimer, age, and fibrinogen degradation products [ 97 ].
Jin et al. reported that only linear discriminant analysis (AUC 0.773) and logistic regression (AUC 0.772) outperformed the Khorana score (AUC 0.642) in cancer-related VTE prediction [ 102 ]. The combination with D-dimer improved the models’ performance [ 103 ]. The top five predictors of cancer-related VTE were D-dimer level, age, Charlson Comorbidity Index, length of stay, and previous VTE history [ 102 ].
Lei et al. recommended the random forest model as the best classifier for VTE prediction in lung cancer [ 103 ]. The model presented an AUC of 0.91 (95% CI: 0.893–0.926), a sensitivity of 0.714 (95% CI: 0.614–0.762), and a specificity of 0.965 (95% CI: 0.941–0.985) [ 103 ]. The five most relevant parameters were Karnofsky Performance Status, a history of VTE, recombinant human endostatin, EGFR-TKI, and platelet count [ 103 ].
Mantha et al. (Preprint) [ 104 ] conducted the first study of a deep-learning model that predicts the risk of cancer-associated VTE. The model was selected based on its C-index and potential usefulness in clinical practice [ 104 ]. The DeepHit model’s most important predictors were plasma albumin, followed by the presence of metastatic disease [ 104 ]. Additionally, the use of systemic therapy, plasma electrolytes (sodium, potassium, chloride, and calcium), hemoglobin, glucose, and alkaline phosphatase were identified as VTE risk predictors [ 104 ]. Meng et al. found in hospitalized cancer patients that the extreme gradient boosting (XGBoost) model achieved the best performance in VTE prediction [ 105 ]. The five most significant features tested in the model were D-dimer level, diabetes, hypertension, pleural metastasis, and hematological malignancies [ 105 ]. Danilatou and colleagues demonstrated that the machine learning approach outperformed traditional scoring systems in predicting early and late mortality in critically ill patients with venous thromboembolism and cancer. In addition, they validated the model externally [ 106 ].
In the future, machine learning models have potential in clinical practice but require optimization with larger databases and multiple algorithms. It is crucial to adhere to the standardization of reporting provided by the Scientific and Standardization Committee (SSC) Subcommittee on Hemostasis & Malignancy of the International Society on Thrombosis and Hemostasis (ISTH). This involves the TRIPOD checklist (Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis), clearly defining predictors, defining the derivation population, and validating the model externally before implementing it [ 107 ].
This paper aims to present the current relevant knowledge in VTE risk assessment in ambulatory cancer patients, starting from the guideline recommendations and continuing with the specific risk assessment methods and machine learning models approaches. The main limitation of this review is its narrative character.
7. Conclusions and Future Directions
VTE risk assessment in ambulatory cancer patients is still challenging. High-risk cancer patients must be accurately discriminated against thromboprophylaxis, but the guidelines do not provide enough information. Many scores, nomograms, and models were developed, but none have optimally performed in this setting. Clinical features, biomarkers, and genetic patterns have been tested alone or grouped in cancer populations in general or specific cancer cohorts. The polygenic risk scores that do not change during the cancer course could be potential predictors of cancer-associated VTE independent of cancer type, but this idea needs further validation in prospective studies. Additionally, the expenses must be assessed better. The machine learning models might provide a potentially useful algorithm through learning and improving its performance based on the data they use. But, to apply these methods in clinical practice, they need to be optimized in larger databases.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.
Author Contributions
Conceptualization, A.D. and A.Ş.D.; methodology, A.D. and A.Ş.D.; validation, A.D. and A.Ş.D.; formal analysis A.D. and A.Ş.D.; investigation A.D. and A.Ş.D.; resources, A.D. and A.Ş.D.; data curation, A.D. and A.Ş.D.; writing—original draft preparation, A.D. and A.Ş.D.; writing—review and editing, A.D. and A.Ş.D.; visualization, A.D. and A.Ş.D.; supervision, A.D. and A.Ş.D.; project administration, A.D. and A.Ş.D.; funding acquisition, A.D. All authors have read and agreed to the published version of the manuscript.
Conflicts of Interest
The authors declare no conflicts of interest.
References
1. Abdol Razak N.B. Jones G. Bhandari M. Berndt M.C. Metharom P. Cancer-Associated Thrombosis: An Overview of Mechanisms, Risk Factors, and Treatment Cancers 2018 10 380 10.3390/cancers10100380 30314362 PMC6209883
2. Mulder F.I. Horváth-Puhó E. van Es N. van Laarhoven H.W.M. Pedersen L. Moik F. Ay C. Büller H.R. Sørensen H.T. Venous thromboembolism in cancer patients: A population-based cohort study Blood 2021 137 1959 1969 10.1182/blood.2020007338 33171494
3. Sørensen H.T. Pedersen L. van Es N. Büller H.R. Horváth-Puhó E. Impact of venous thromboembolism on the mortality in patients with cancer: A population-based cohort study Lancet Reg. Health Eur. 2023 34 100739 10.1016/j.lanepe.2023.100739 37809052 PMC10558815
4. Khorana A.A. McCrae K.R. Milentijevic D. Laliberté F. Lejeune D. Crivera C. Lefebvre P. Healthcare resource utilization and costs associated with venous thromboembolism recurrence in patients with cancer J. Med. Econ. 2020 23 323 329 10.1080/13696998.2019.1703190 31818164
5. Falanga A. Marchetti M. Cancer-associated thrombosis: Enhanced awareness and pathophysiologic complexity J. Thromb. Haemost. 2023 21 1397 1408 10.1016/j.jtha.2023.02.029 36931602
6. Falanga A. Ay C. Di Nisio M. Gerotziafas G. Jara-Palomares L. Langer F. Lecumberri R. Mandala M. Maraveyas A. Pabinger I. Venous thromboembolism in cancer patients: ESMO Clinical Practice Guideline Ann. Oncol. 2023 34 452 467 10.1016/j.annonc.2022.12.014 36638869
7. Key N.S. Khorana A.A. Kuderer N.M. Bohlke K. Lee A.Y.Y. Arcelus J.I. Wong S.L. Balaban E.P. Flowers C.R. Gates L.E. Venous Thromboembolism Prophylaxis and Treatment in Patients with Cancer: ASCO Guideline Update J. Clin. Oncol. 2023 41 3063 3071 10.1200/JCO.23.00294 37075273
8. Lyman G.H. Carrier M. Ay C. Di Nisio M. Hicks L.K. Khorana A.A. Leavitt A.D. Lee A.Y.Y. Macbeth F. Morgan R.L. American Society of Hematology 2021 guidelines for management of venous thromboembolism: Prevention and treatment in patients with cancer Blood Adv. 2021 5 927 974 10.1182/bloodadvances.2020003442 33570602 PMC7903232
9. Middeldorp S. Nieuwlaat R. Baumann Kreuziger L. Coppens M. Houghton D.E. James A.H. Lang E. Moll S. Myers T. Bhatt M. American Society of Hematology 2023 Guidelines for Management of Venous Thromboembolism: Thrombophilia Testing Blood Adv. 2023 7 7101 7138 10.1182/bloodadvances.2023010177 37195076 PMC10709681
10. Lyon A.R. López-Fernández T. Couch L.S. Asteggiano R. Aznar M.C. Bergler-Klein J. Boriani G. Cardinale D. Cordoba R. Cosyns B. 2022 ESC Guidelines on cardio-oncology developed in collaboration with the European Hematology Association (EHA), the European Society for Therapeutic Radiology and Oncology (ESTRO) and the International Cardio-Oncology Society (IC-OS) Eur. Heart J. 2022 43 4229 4361 10.1093/eurheartj/ehac244 36017568
11. Drăgan A. Sinescu I. The Role of the Cardiac Biomarkers in the Renal Cell Carcinoma Multidisciplinary Management Diagnostics 2023 13 1912 10.3390/diagnostics13111912 37296764 PMC10253077
12. Mosarla R.C. Vaduganathan M. Qamar A. Moslehi J. Piazza G. Giugliano R.P. Anticoagulation Strategies in Patients with Cancer: JACC Review Topic of the Week J. Am. Coll. Cardiol. 2019 73 1336 1349 10.1016/j.jacc.2019.01.017 30898209 PMC7957366
13. van Es N. Ay C. Jara-Palomares L. Screening for Occult Cancer in Patients with Venous Thromboembolism: Past, Present, and Future Hamostaseologie 2020 40 270 279 10.1055/a-1150-2286 32726822
14. Gainsbury M.L. Erdrich J. Taubman D. Mirocha J. Manguso N. Amersi F. Silberman A.W. Prevalence and Predictors of Preoperative Venous Thromboembolism in Asymptomatic Patients Undergoing Major Oncologic Surgery Ann. Surg. Oncol. 2018 25 1640 1645 10.1245/s10434-018-6461-2 29626305
15. Heidrich H. Konau E. Hesse P. Asymptomatic venous thrombosis in cancer patients—A problem often overlooked. Results of a retrospective and prospective study Vasa 2009 38 160 166 10.1024/0301-1526.38.2.160 19588304
16. Loftus J.R. Hu Z. Morin B.R. Hobbs S.K. Francis C.W. Khorana A.A. Rubens D.J. Kaproth-Joslin K.A. Vascular Imaging in the Asymptomatic High-risk Cancer Population: A Role for Thrombosis Screening and Therapy Management J. Ultrasound Med. 2022 41 225 236 10.1002/jum.15701 33772825
17. Kourlaba G. Gourzoulidis G. Rasmussen E. Kontodimas S. Maniadakis N. Cost Effectiveness of Ultrasound Screening, Cancer Patients, to Detect Asymptomatic Thrombosis Value Health 2017 20 A587 10.1016/j.jval.2017.08.1069
18. Kunapareddy G. Switzer B. Jain P. Conces M. Chen Y.W. Patel B. Patel S. Pinnamaneni P. Pohlman B. Angelini D.E. Implementation of an electronic medical record tool for early detection of deep vein thrombosis in the ambulatory oncology setting Res. Pract. Thromb. Haemost. 2019 3 226 233 10.1002/rth2.12176 31011706 PMC6462749
19. Holmes C.E. Ades S. Gilchrist S. Douce D. Libby K. Rogala B. Parenteau E. Cushman M. Holm A.K. Successful Model for Guideline Implementation to Prevent Cancer-Associated Thrombosis: Venous Thromboembolism Prevention in the Ambulatory Cancer Clinic JCO Oncol. Pract. 2020 16 e868 e874 10.1200/JOP.19.00697 32267798 PMC7489490
20. Ay C. Vormittag R. Dunkler D. Simanek R. Chiriac A.L. Drach J. Quehenberger P. Wagner O. Zielinski C. Pabinger I. 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 4124 4129 10.1200/JCO.2008.21.7752 19636003
21. Schorling R.M. Pfrepper C. Golombek T. Cella C.A. Muñoz-Unceta N. Siegemund R. Engel C. Petros S. Lordick F. Knödler M. Evaluation of Biomarkers for the Prediction of Venous Thromboembolism in Ambulatory Cancer Patients Oncol. Res. Treat. 2020 43 414 427 10.1159/000508271 32580190
22. Niimi K. Nishida K. Lee C. Ikeda S. Kawai Y. Sugimoto M. Banno H. Optimal D-Dimer Cutoff Values for Diagnosing Deep Vein Thrombosis in Patients with Comorbid Malignancies Ann. Vasc. Surg. 2024 98 293 300 10.1016/j.avsg.2023.06.033 37454901
23. Alexander M. Harris S. Underhill C. Torres J. Sharma S. Lee N. Wong H. Eek R. Michael M. Tie J. Risk-Directed Ambulatory Thromboprophylaxis in Lung and Gastrointestinal Cancers: The TARGET-TP Randomized Clinical Trial JAMA Oncol. 2023 9 1536 1545 10.1001/jamaoncol.2023.3634 37733336 PMC10514890
24. Oi M. Yamashita Y. Toyofuku M. Morimoto T. Motohashi Y. Tamura T. Kaitani K. Amano H. Takase T. Hiramori S. D-dimer levels at diagnosis and long-term clinical outcomes in venous thromboembolism: From the COMMAND VTE Registry J. Thromb. Thrombolysis. 2020 49 551 561 10.1007/s11239-019-01964-0 31571121
25. Koch V. Martin S.S. Gruber-Rouh T. Eichler K. Mahmoudi S. Leistner D.M. Scholtz J.E. Bernatz S. Puntmann V.O. Nagel E. Cancer patients with venous thromboembolism: Diagnostic and prognostic value of elevated D-dimers Eur. J. Clin. Investig. 2023 53 e13914 10.1111/eci.13914 36444723
26. Ay C. Simanek R. Vormittag R. Dunkler D. Alguel G. Koder S. Kornek G. Marosi C. Wagner O. Zielinski C. High plasma levels of soluble P-selectin are predictive of venous thromboembolism in cancer patients: Results from the Vienna Cancer and Thrombosis Study (CATS) Blood 2008 112 2703 2708 10.1182/blood-2008-02-142422 18539899
27. Zhang X. Zhang C. Ma Z. Zhang Y. Soluble P-selectin level in patients with cancer-associated venous and artery thromboembolism: A systematic review and meta-analysis Arch. Med. Sci. 2023 19 274 282 10.5114/aoms/159039 36817657 PMC9897104
28. Khorana A.A. Barnard J. Wun T. Vijapurkar U. Damaraju C.V. Moore K.T. Wildgoose P. McCrae K.R. Biomarker signatures in cancer patients with and without venous thromboembolism events: A substudy of CASSINI Blood Adv. 2022 6 1212 1221 10.1182/bloodadvances.2021005710 34807979 PMC8864658
29. Anijs R.J.S. Nguyen Y.N. Cannegieter S.C. Versteeg H.H. Buijs J.T. MicroRNAs as prognostic biomarkers for (cancer-associated) venous thromboembolism J. Thromb. Haemost. 2023 21 7 17 10.1016/j.jtha.2022.09.001 36695398
30. Marques I.S. Tavares V. Neto B.V. Mota I.N.R. Pereira D. Medeiros R. Long Non-Coding RNAs in Venous Thromboembolism: Where Do We Stand? Int. J. Mol. Sci. 2023 24 12103 10.3390/ijms241512103 37569483 PMC10418965
31. Khorana A.A. Kuderer N.M. Culakova E. Lyman G.H. Francis C.W. Development and validation of a predictive model for chemotherapy-associated thrombosis Blood 2008 111 4902 4907 10.1182/blood-2007-10-116327 18216292 PMC2384124
32. Khorana A.A. DeSancho M.T. Liebman H. Rosovsky R. Connors J.M. Zwicker J. Prediction and Prevention of Cancer-Associated Thromboembolism Oncologist 2021 26 e2 e7 10.1002/onco.13569 33274815 PMC7794203
33. Mulder F.I. Candeloro M. Kamphuisen P.W. Di Nisio M. Bossuyt P.M. Guman N. Smit K. Buller H.R. van Es N. CAT-prediction collaborators The Khorana score for prediction of venous thromboembolism in cancer patients: A systematic review and meta-analysis Haematologica 2019 104 1277 1287 10.3324/haematol.2018.209114 30606788 PMC6545838
34. Akasaka-Kihara F. Sueta D. Ishii M. Maki Y. Hirakawa K. Tabata N. Ito M. Yamanaga K. Fujisue K. Hoshiyama T. Validation of the Khorana Venous Thromboembolism Risk Score in Japanese Cancer Patients JACC Asia 2021 1 259 270 10.1016/j.jacasi.2021.07.006 36338156 PMC9627826
35. Ramos-Esquivel A. Marenco-Flores A. Hernández-Romero G. Umaña-Mora C. Céspedes-Calvo A. Mora-Hidalgo R. Using the Khorana risk score to predict venous thromboembolism and overall survival in a cohort of Hispanic patients with solid malignancies Ecancermedicalscience 2022 16 1470 10.3332/ecancer.2022.1470 36819798 PMC9934874
36. El-Sayed H.A. Othman M. Azzam H. Bucciol R. Ebrahim M.A. El-Agdar M.A.M.A. Tera Y. Sakr D.H. Ghoneim H.R. Selim T.E. Assessing the risk of venous thromboembolism in patients with haematological cancers using three prediction models J. Cancer Res. Clin. Oncol. 2023 149 17771 17780 10.1007/s00432-023-05475-7 37935936 PMC11797160
37. Ha H. Ko Y.H. Kim K. Hong J. Lee G.W. Jeong S.H. Bang S.M. Yoon S.S. Application of the Khorana score for cancer-associated thrombosis prediction in patients of East Asian ethnicity undergoing ambulatory chemotherapy Thromb. J. 2023 21 63 10.1186/s12959-023-00505-3 37271814 PMC10240689
38. Overvad T.F. Ording A.G. Nielsen P.B. Skjøth F. Albertsen I.E. Noble S. Vistisen A.K. Gade I.L. Severinsen M.T. Piazza G. Validation of the Khorana score for predicting venous thromboembolism in 40 218 patients with cancer initiating chemotherapy Blood Adv. 2022 6 2967 2976 10.1182/bloodadvances.2021006484 35045569 PMC9131922
39. van Es N. Ventresca M. Di Nisio M. Zhou Q. Noble S. Crowther M. Briel M. Garcia D. Lyman G.H. Macbeth F. IPDMA Heparin Use in Cancer Patients Research Group. The Khorana score for prediction of venous thromboembolism in cancer patients: An individual patient data meta-analysis J. Thromb. Haemost. 2020 18 1940 1951 10.1111/jth.14824 32336010
40. Austin K. George J. Robinson E.J. Scully M. Thomas M.R. Retrospective Cohort Study of Venous Thromboembolism Rates in Ambulatory Cancer Patients: Association with Khorana Score and Other Risk Factors J. Hematol. 2019 8 17 25 10.14740/jh471 32300436 PMC7153676
41. Verzeroli C. Giaccherini C. Russo L. Bolognini S. Gamba S. Tartari C.J. Schieppati F. Ticozzi C. Vignoli A. Masci G. Utility of the Khorana and the new-Vienna CATS prediction scores in cancer patients of the HYPERCAN cohort J. Thromb. Haemost. 2023 21 1869 1881 10.1016/j.jtha.2023.03.037 37054917
42. Abdel-Razeq H. Sharaf B. Al-Jaghbeer M.J. Abu-Fares H. Bater R. Shaer M.A. Abu-Jaish H. Laban D.A. Salamah O. Tamimi F. COMPASS-CAT versus Khorana risk assessment model for predicting venous thromboembolic events in patients with non-small cell lung cancer on active treatment with chemotherapy and/or immunotherapy, the CK-RAM study J. Thromb. Thrombolysis 2023 56 447 453 10.1007/s11239-023-02860-4 37430158
43. Mansfield A.S. Tafur A.J. Wang C.E. Kourelis T.V. Wysokinska E.M. Yang P. Predictors of active cancer thromboembolic outcomes: Validation of the Khorana score among patients with lung cancer J. Thromb. Haemost. 2016 14 1773 1778 10.1111/jth.13378 27273134 PMC5035574
44. Sanfilippo K.M. Carson K.R. Wang T.F. Luo S. Edwin N. Kuderer N. Keller J.M. Gage B.F. Evaluation of the Khorana score for prediction of venous thromboembolism in patients with multiple myeloma Res. Pract. Thromb. Haemost. 2022 6 e12634 10.1002/rth2.12634 35028491 PMC8742966
45. Mirza A.S. Yun S. Ali N.A. Shin H. O’Neil J.L. Elharake M. Schwartz D. Robinson K. Nowell E. Engle G. Validation of the Khorana score in acute myeloid leukemia patients: A single-institution experience Thromb. J. 2019 17 13 10.1186/s12959-019-0202-z 31303864 PMC6604148
46. Wang Y. Attar B.M. Fuentes H.E. Yu J. Zhang H. Tafur A.J. Performance of Khorana Risk Score for Prediction of Venous Thromboembolism in Patients with Hepatocellular Carcinoma Clin. Appl. Thromb. Hemost. 2018 24 471 476 10.1177/1076029617699088 28288526 PMC6714645
47. Piver R.N. Wagner V.M. Levine M.D. Backes F.J. Chambers L.J. Cohn D.E. Copeland L.J. Cosgrove C.M. Nagel C.I. O’Malley D.M. Use of the Khorana score to predict venous thromboembolism in patients undergoing chemotherapy for uterine cancer Gynecol. Oncol. Rep. 2023 46 101156 10.1016/j.gore.2023.101156 36910448 PMC9995928
48. Rupa-Matysek J. Gil L. Kaźmierczak M. Barańska M. Komarnicki M. Prediction of venous thromboembolism in newly diagnosed patients treated for lymphoid malignancies: Validation of the Khorana Risk Score Med. Oncol. 2017 35 5 10.1007/s12032-017-1065-4 29209847 PMC5717131
49. Di Nisio M. van Es N. Rotunno L. Anzoletti N. Falcone L. De Tursi M. Natoli C. Tinari N. Cavallo I. Valeriani E. Long-term performance of risk scores for venous thromboembolism in ambulatory cancer patients J. Thromb. Thrombolysis 2019 48 125 133 10.1007/s11239-019-01845-6 30919253
50. Guman N.A.M. van Geffen R.J. Mulder F.I. van Haaps T.F. Hovsepjan V. Labots M. Cirkel G.A. de Vos F.Y.F.L. Ten Tije A.J. Beerepoot L.V. Evaluation of the Khorana, PROTECHT, and 5-SNP scores for prediction of venous thromboembolism in patients with cancer J. Thromb. Haemost. 2021 19 2974 2983 10.1111/jth.15503 34409743 PMC9291564
51. Ay C. Dunkler D. Marosi C. Chiriac A.L. Vormittag R. Simanek R. Quehenberger P. Zielinski C. Pabinger I. Prediction of venous thromboembolism in cancer patients Blood 2010 116 5377 5382 10.1182/blood-2010-02-270116 20829374
52. van Es N. Di Nisio M. Cesarman G. Kleinjan A. Otten H.M. Mahé I. Wilts I.T. Twint D.C. Porreca E. Arrieta O. Comparison of risk prediction scores for venous thromboembolism in cancer patients: A prospective cohort study Haematologica 2017 102 1494 1501 10.3324/haematol.2017.169060 28550192 PMC5685240
53. Harada Y. Sato A. Nishioka A. Ogusu S. Matsumoto M. Sueoka E. Kawaguchi A. Kimura S. Sueoka-Aragane N. Usefulness of blood biomarkers for predicting venous thromboembolism in Japanese patients with cancer Oncol. Lett. 2023 25 180 10.3892/ol.2023.13766 37033099 PMC10079810
54. 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 7 291 292 10.1007/s11739-012-0784-y 22547369
55. Moik F. van Es N. Posch F. Di Nisio M. Fuereder T. Preusser M. Pabinger I. Ay C. Gemcitabine and Platinum-Based Agents for the Prediction of Cancer-Associated Venous Thromboembolism: Results from the Vienna Cancer and Thrombosis Study Cancers 2020 12 2493 10.3390/cancers12092493 32899157 PMC7564761
56. Ramos-Esquivel A. Marenco-Flores A. Hernández-Romero G. Céspedes-Calvo A. Mora-Hidalgo R. Comparison among three predictive models for cancer-associated thromboembolism in a hispanic population J. Thromb. Thrombolysis 2023 56 433 438 10.1007/s11239-023-02861-3 37407771
57. Pelzer U. Sinn M. Stieler J. Riess H. Primary pharmacological prevention of thromboembolic events in ambulatory patients with advanced pancreatic cancer treated with chemotherapy? Dtsch. Med. Wochenschr. 2013 138 2084 2088 24085361 10.1055/s-0033-1349608
58. Qin L. Liang Z. Xie J. Li X. Estimating Venous Thromboembolism Risk in Metastatic Colorectal Cancer Inpatients: Validation of Existing Risk Scores and Development of New Risk Scores Clin. Appl. Thromb. Hemost. 2023 29 10760296231196859 10.1177/10760296231196859 37691565 PMC10498692
59. Gomez-Rosas P. Giaccherini C. Russo L. Verzeroli C. Gamba S. Tartari C.J. Bolognini S. Ticozzi C. Schieppati F. Barcella L. A New Risk Prediction Model for Venous Thromboembolism and Death in Ambulatory Lung Cancer Patients Cancers 2023 15 4588 10.3390/cancers15184588 37760562 PMC10527104
60. Yan A.R. Samarawickrema I. Naunton M. Peterson G.M. Yip D. Mortazavi R. Models for predicting venous thromboembolism in ambulatory patients with lung cancer: A systematic review protocol BMJ Open 2021 11 e055322 10.1136/bmjopen-2021-055322 PMC8638451 34853112
61. Pabinger I. van Es N. Heinze G. Posch F. Riedl J. Reitter E.M. Di Nisio M. Cesarman-Maus G. Kraaijpoel N. Zielinski C.C. A clinical prediction model for cancer-associated venous thromboembolism: A development and validation study in two independent prospective cohorts Lancet Haematol. 2018 5 e289 e298 10.1016/S2352-3026(18)30063-2 29885940 PMC7338218
62. Gerotziafas G.T. Taher A. Abdel-Razeq H. AboElnazar E. Spyropoulos A.C. El Shemmari S. Larsen A.K. Elalamy I. 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 1222 1231 10.1634/theoncologist.2016-0414 28550032 PMC5634762
63. Taglialatela I. Mariani L. Dotti K.F. Di Vico L. Pisanu M.N. Facchinetti C. De Braud F. Ferrari L.A.M. Central venous catheters-related-thrombosis and risk factors in oncological patients: A retrospective evaluation of recent risk scores Tumori J. 2023 109 363 369 10.1177/03008916221111419 35815563 PMC10363937
64. Rupa-Matysek J. Lembicz M. Rogowska E.K. Gil L. Komarnicki M. Batura-Gabryel H. Evaluation of risk factors and assessment models for predicting venous thromboembolism in lung cancer patients Med. Oncol. 2018 35 63 10.1007/s12032-018-1120-9 29616356 PMC5882764
65. Spyropoulos A.C. Eldredge J.B. Anand L.N. Zhang M. Qiu M. Nourabadi S. Rosenberg D.J. External Validation of a Venous Thromboembolic Risk Score for Cancer Outpatients with Solid Tumors: The COMPASS-CAT Venous Thromboembolism Risk Assessment Model Oncologist 2020 25 e1083 e1090 10.1634/theoncologist.2019-0482 32017293 PMC7356707
66. Pestana R.M.C. Alves M.T. de Oliveira A.N. Oliveira H.H.M. Soares C.E. Sabino A.P. Silva L.M. Simões R. Gomes K.B. Interleukin-10 levels and the risk of thromboembolism according to COMPASS-Cancer associated thrombosis score in breast cancer patients prior to undergoing doxorubicin-based chemotherapy Blood Coagul. Fibrinolysis 2023 34 70 74 10.1097/MBC.0000000000001159 35946469
67. Cella C.A. Di Minno G. Carlomagno C. Arcopinto M. Cerbone A.M. Matano E. Tufano A. Lordick F. De Simone B. Muehlberg K.S. Preventing Venous Thromboembolism in Ambulatory Cancer Patients: The ONKOTEV Study Oncologist 2017 22 601 608 10.1634/theoncologist.2016-0246 28424324 PMC5423517
68. Cella C.A. Knoedler M. Hall M. Arcopinto M. Bagnardi V. Gervaso L. Pellicori S. Spada F. Zampino M.G. Ravenda P.S. Validation of the ONKOTEV Risk Prediction Model for Venous Thromboembolism in Outpatients with Cancer JAMA Netw. Open 2023 6 e230010 10.1001/jamanetworkopen.2023.0010 36795409 PMC9936336
69. Godinho J. Casa-Nova M. Moreira-Pinto J. Simões P. Paralta Branco F. Leal-Costa L. Faria A. Lopes F. Teixeira J.A. Passos-Coelho J.L. ONKOTEV Score as a Predictive Tool for Thromboembolic Events in Pancreatic Cancer—A Retrospective Analysis Oncologist 2020 25 e284 e290 10.1634/theoncologist.2019-0510 32043787 PMC7011666
70. Xiong W. Zhao Y. Du H. Wang Y. Xu M. Guo X. Optimal authoritative risk assessment score of Cancer-associated venous thromboembolism for hospitalized medical patients with lung Cancer Thromb. J. 2021 19 95 10.1186/s12959-021-00339-x 34863189 PMC8642841
71. Pfrepper C. Knödler M. Schorling R.M. Seehofer D. Petros S. Lordick F. Predictors for thromboembolism in patients with cholangiocarcinoma J. Cancer Res. Clin. Oncol. 2022 148 2415 2426 10.1007/s00432-021-03794-1 34499224 PMC9349130
72. Zhang L. Fang Y. Xing J. Cheng H. Sun X. Yuan Z. Xu Y. Hao J. The Efficacy of the Systemic Immune-Inflammation Index and Prognosis Nutritional Index for the Diagnosis of Venous Thromboembolism in Gastrointestinal Cancers J. Inflamm. Res. 2022 15 4649 4661 10.2147/JIR.S376601 35996687 PMC9391990
73. Tort M. Sevil F.C. Sevil H. Becit N. Evaluation of systemic immune-inflammation index in acute deep vein thrombosis: A propensity-matched J. Vasc. Surg. Venous Lymphat. Disord. 2023 11 972 977.e1 10.1016/j.jvsv.2023.02.008 37003462
74. Zhang L. Liu X. Yang R. Yang Y. Chen X. The Diagnostic Value of the Systemic Immune-Inflammation Index for Venous Thromboembolism in Lung Cancer Patients: A Retrospective Study Mediat. Inflamm. 2022 2022 9215311 10.1155/2022/9215311 36046761 PMC9424048
75. Li A. La J. May S.B. Guffey D. da Costa W.L. Jr. Amos C.I. Bandyo R. Milner E.M. Kurian K.M. Chen D.C.R. Derivation and Validation of a Clinical Risk Assessment Model for Cancer-Associated Thrombosis in Two Unique US Health Care Systems J. Clin. Oncol. 2023 41 2926 2938 10.1200/JCO.22.01542 36626707 PMC10431461
76. Castellón Rubio V.E. Segura P.P. Muñoz A. Farré A.L. Ruiz L.C. Lorente J. High plasma levels of soluble P-Selectin and Factor VIII predict venous thromboembolism in non-small cell lung cancer patients: The Thrombo-Nsclc risk score Thromb. Res. 2020 196 349 354 10.1016/j.thromres.2020.09.021 32977135
77. Li H. Tian Y. Niu H. He L. Cao G. Zhang C. Kaiweisierkezi K. Luo Q. Derivation, validation and assessment of a novel nomogram-based risk assessment model for venous thromboembolism in hospitalized patients with lung cancer: A retrospective case control study Front. Oncol. 2022 12 988287 10.3389/fonc.2022.988287 36300098 PMC9589115
78. Antic D. Milic N. Nikolovski S. Todorovic M. Bila J. Djurdjevic P. Andjelic B. Djurasinovic V. Sretenovic A. Vukovic V. Development and validation of multivariable predictive model for thromboembolic events in lymphoma patients Am. J. Hematol. 2016 91 1014 1019 10.1002/ajh.24466 27380861
79. Kirkizlar O. Alp Kirkizlar T. Umit E.G. Asker I. Baysal M. Bas V. Gulsaran S.K. Demirci U. Demir A.M. The Incidence of Venous Thromboembolism and Impact on Survival in Hodgkin Lymphoma Clin. Lymphoma Myeloma Leuk. 2020 20 542 547 10.1016/j.clml.2020.02.021 32245743
80. Abdel-Razeq H. Ma’koseh M. Mansour A. Bater R. Amarin R. Abufara A. Halahleh K. Manassra M. Alrwashdeh M. Almomani M. The Application of the ThroLy Risk Assessment Model to Predict Venous Thromboembolism in Patients with Diffuse Large B-Cell Lymphoma Clin. Appl. Thromb. Hemost. 2021 27 10760296211045908 10.1177/10760296211045908 34590497 PMC8642105
81. Rupa-Matysek J. Brzeźniakiewicz-Janus K. Gil L. Krasiński Z. Komarnicki M. Evaluation of the ThroLy score for the prediction of venous thromboembolism in newly diagnosed patients treated for lymphoid malignancies in clinical practice Cancer Med. 2018 7 2868 2875 10.1002/cam4.1540 29761831 PMC6051175
82. Bastos-Oreiro M. Ortiz J. Pradillo V. Salas E. Marínez-Laperche C. Muñoz A. Buño I. Diéz-Martin J.L. Soria J.M. Pascual Izquierdo C. Incorporating genetic and clinical data into the prediction of thromboembolism risk in patients with lymphoma Cancer Med. 2021 10 7585 7592 10.1002/cam4.4280 34598309 PMC8559493
83. López Sacerio A. Tejeda Ramón M.C. Morales Helguera A. Pérez Castillo Y. Cruz Rodríguez J. Guerra Rodríguez J.F. Falanga A. Validation of venous thromboembolism predictive model in hematologic malignancies Ann. Hematol. 2023 102 3613 3620 10.1007/s00277-023-05463-4 37782372
84. Sanfilippo K.M. Luo S. Wang T.F. Fiala M. Schoen M. Wildes T.M. Mikhael J. Kuderer N.M. Calverley D.C. Keller J. Predicting venous thromboembolism in multiple myeloma: Development and validation of the IMPEDE VTE score Am. J. Hematol. 2019 94 1176 1184 10.1002/ajh.25603 31379000 PMC7058359
85. Sanfilippo K.M. Fiala M. Feinberg D. Tathireddy H. Girard T. Vij R. Di Paola J. Gage B.F. D-dimer predicts venous thromboembolism in multiple myeloma: A nested case-control study Res. Pract. Thromb. Haemost. 2023 7 e102235 10.1016/j.rpth.2023.102235 38193055 PMC10772878
86. Wolpert F. Berghoff A.S. Grossenbacher B. Lareida A. Löb R. Roth P. Neidert M.C. Andratschke N. Le Rhun E. Preusser M. Venous thromboembolic events in patients with brain metastases: The PICOS score Eur. J. Cancer 2020 134 75 85 10.1016/j.ejca.2020.04.018 32473541
87. Kubo K. Nakamura K. Okamoto K. Matsuoka H. Ida N. Haruma T. Ogawa C. Masuyama H. The Combination of D-dimer and Glasgow Prognostic Score Can Be Useful in Predicting VTE in Patients with Stage IIIC and IVA Ovarian Cancer Acta Med. Okayama 2022 76 129 135 35503440 10.18926/AMO/63406
88. Roy D.C. Wang T.F. Carrier M. Mallick R. Burger D. Hawken S. Wells P.S. Thrombophilia gene mutations predict venous thromboembolism in ambulatory cancer patients receiving chemotherapy J. Thromb. Haemost. 2023 21 3184 3192 10.1016/j.jtha.2023.07.017 37536569
89. Lindström S. Wang L. Smith E.N. Gordon W. van Hylckama Vlieg A. de Andrade M. Brody J.A. Pattee J.W. Haessler J. Brumpton B.M. Genomic and transcriptomic association studies identify 16 novel susceptibility loci for venous thromboembolism Blood 2019 134 1645 1657 10.1182/blood.2019000435 31420334 PMC6871304
90. de Haan H.G. Bezemer I.D. Doggen C.J. Le Cessie S. Reitsma P.H. Arellano A.R. Tong C.H. Devlin J.J. Bare L.A. Rosendaal F.R. Multiple SNP testing improves risk prediction of first venous thrombosis Blood 2012 120 656 663 10.1182/blood-2011-12-397752 22586183
91. Skille H. Paulsen B. Hveem K. Gabrielsen M.E. Brumpton B. Hindberg K. Gran O.V. Rosendaal F.R. Braekkan S.K. Hansen J.B. Combined effects of five prothrombotic genotypes and cancer on the risk of a first venous thromboembolic event J. Thromb. Haemost. 2020 18 2861 2869 10.1111/jth.15011 32671915
92. Jakobsen L. Frischmuth T. Brækkan S.K. Hansen J.B. Morelli V.M. Joint Effect of Multiple Prothrombotic Genotypes and Mean Platelet Volume on the Risk of Incident Venous Thromboembolism Thromb. Haemost. 2022 122 1911 1920 10.1055/a-1863-2052 35617954
93. Guman N.A.M. Mulder F.I. Ferwerda B. Zwinderman A.H. Kamphuisen P.W. Büller H.R. van Es N. Polygenic risk scores for prediction of cancer-associated venous thromboembolism in the UK Biobank cohort study J. Thromb. Haemost. 2023 21 3175 3183 10.1016/j.jtha.2023.07.009 37481074
94. Muñoz Martín A.J. Ortega I. Font C. Pachón V. Castellón V. Martínez-Marín V. Salgado M. Martínez E. Calzas J. Rupérez A. Multivariable clinical-genetic risk model for predicting venous thromboembolic events in patients with cancer Br. J. Cancer 2018 118 1056 1061 10.1038/s41416-018-0027-8 29588512 PMC5931103
95. Neto B.V. Tavares V. da Silva J.B. Liz-Pimenta J. Marques I.S. Carvalho L. Salgado L. Pereira D. Medeiros R. Thrombogenesis-associated genetic determinants as predictors of thromboembolism and prognosis in cervical cancer Sci. Rep. 2023 13 9519 10.1038/s41598-023-36161-w 37308506 PMC10260924
96. Muñoz A. Ay C. Grilz E. López S. Font C. Pachón V. Castellón V. Martínez-Marín V. Salgado M. Martínez E. A Clinical-Genetic Risk Score for Predicting Cancer-Associated Venous Thromboembolism: A Development and Validation Study Involving Two Independent Prospective Cohorts J. Clin. Oncol. 2023 41 2911 2925 10.1200/JCO.22.00255 36730884 PMC10414737
97. Xu Q. Lei H. Li X. Li F. Shi H. Wang G. Sun A. Wang Y. Peng B. Machine learning predicts cancer-associated venous thromboembolism using clinically available variables in gastric cancer patients Heliyon 2023 9 e12681 10.1016/j.heliyon.2022.e12681 36632097 PMC9826862
98. Nudel J. Bishara A.M. de Geus S.W.L. Patil P. Srinivasan J. Hess D.T. Woodson J. Development and validation of machine learning models to predict gastrointestinal leak and venous thromboembolism after weight loss surgery: An analysis of the MBSAQIP database Surg. Endosc. 2021 35 182 191 10.1007/s00464-020-07378-x 31953733 PMC9278895
99. Ting Sim J.Z. Fong Q.W. Huang W. Tan C.H. Machine learning in medicine: What clinicians should know Singap. Med. J. 2023 64 91 97 10.11622/smedj.2021054 PMC10071847 34005847
100. Ferroni P. Zanzotto F.M. Scarpato N. Riondino S. Nanni U. Roselli M. Guadagni F. Risk Assessment for Venous Thromboembolism in Chemotherapy-Treated Ambulatory Cancer Patients Med. Decis. Mak. 2017 37 234 242 10.1177/0272989X16662654 27491558
101. Ferroni P. Zanzotto F.M. Scarpato N. Riondino S. Guadagni F. Roselli M. Validation of a Machine Learning Approach for Venous Thromboembolism Risk Prediction in Oncology Dis. Markers 2017 2017 8781379 10.1155/2017/8781379 29104344 PMC5623790
102. Jin S. Qin D. Liang B.S. Zhang L.C. Wei X.X. Wang Y.J. Zhuang B. Zhang T. Yang Z.P. Cao Y.W. Machine learning predicts cancer-associated deep vein thrombosis using clinically available variables Int. J. Med. Inform. 2022 161 104733 10.1016/j.ijmedinf.2022.104733 35299099
103. Lei H. Zhang M. Wu Z. Liu C. Li X. Zhou W. Long B. Ma J. Zhang H. Wang Y. Development and Validation of a Risk Prediction Model for Venous Thromboembolism in Lung Cancer Patients Using Machine Learning Front. Cardiovasc. Med. 2022 9 845210 10.3389/fcvm.2022.845210 35321110 PMC8934875
104. Mantha S. Chatterjee S. Singh R. Cadley J. Poon C. Chatterjee A. Kelly D. Sterpi M. Soff G. Zwicker J. Application of Machine Learning to the Prediction of Cancer-Associated Venous Thromboembolism Res. Sq. 2023 rs.3.rs 2870367
105. Meng L. Wei T. Fan R. Su H. Liu J. Wang L. Huang X. Qi Y. Li X. Development and validation of a machine learning model to predict venous thromboembolism among hospitalized cancer patients Asia Pac. J. Oncol. Nurs. 2022 9 100128 10.1016/j.apjon.2022.100128 36276886 PMC9583033
106. Danilatou V. Nikolakakis S. Antonakaki D. Tzagkarakis C. Mavroidis D. Kostoulas T. Ioannidis S. Outcome Prediction in Critically-Ill Patients with Venous Thromboembolism and/or Cancer Using Machine Learning Algorithms: External Validation and Comparison with Scoring Systems Int. J. Mol. Sci. 2022 23 7132 10.3390/ijms23137132 35806137 PMC9266386
107. Sanfilippo K.M. Wang T.F. Carrier M. Falanga A. Gage B.F. Khorana A.A. Maraveyas A. Soff G.A. Wells P.S. Zwicker J.I. Standardization of risk prediction model reporting in cancer-associated thrombosis: Communication from the ISTH SSC subcommittee on hemostasis and malignancy J. Thromb. Haemost. 2022 20 1920 1927 10.1111/jth.15759 35635332