Advances in occult cancer screening for patients with unprovoked venous thromboembolism: A narrative review of epidemiology, risk, and clinical strategies

Abstract

Background

Venous thromboembolism (VTE) and cancer exhibit a bidirectional correlation. The probability of detecting occult cancer in unprovoked VTE patients is significantly increased, and the cancer is often diagnosed at an advanced stage. Early screening is critical for improving prognosis; however, the effectiveness of current risk stratification and screening strategies remains controversial.

Methods

This review systematically integrated evidence on the epidemiology, risk stratification, and screening methods for occult malignancies in individuals with unprovoked VTE.

Results

Cancer-induced hypercoagulability and VTE-related inflammation interact bidirectionally, promoting thrombosis and cancer progression. In terms of risk stratification, elderly patients with VTE, as well as those comorbid with diabetes, diverticular disease, dementia, or with a history of aspirin use, have a higher detection rate of occult cancer. Occult cancer may also be indicated in patients with VTE at special sites, such as splanchnic vein thrombosis, cerebral venous thrombosis, and lower extremity arterial thrombosis, as well as in those with recurrent VTE. However, the impact of sex on the presence of occult cancer in VTE patients remains controversial. International guidelines recommend limited screening as a first-line approach, because extensive screening does not significantly improve prognosis. Positron emission tomography with computed tomography scan may enhance accurate cancer diagnosis. The Registro Informatizado Enfermedad TromboEmbólica (RIETE) and Screening for Occult Malignancy in Patients with Idiopathic Venous Thromboembolism (SOME) risk scores show limited predictive efficacy, while biomarkers and machine learning models demonstrate high diagnostic efficacy, indicating their potential application.

Conclusions

Regular cancer screening is necessary for individuals with unprovoked VTE, and clinical practice should adopt individualized screening strategies based on risk stratification. Future research should focus on optimizing existing models and exploring the combined application of biomarkers and machine learning to improve cancer screening for this population.

Venous thromboembolism (VTE) is a common vascular disease that includes both deep vein thrombosis (DVT) and pulmonary embolism (PE), and it has become a severe public health challenge worldwide. Epidemiological data indicate that it affects >10 million people annually and is associated with significant disability rates and mortality risks.1, 2, 3 Provoked VTE is typically defined as VTE occurring in the presence of risk factors, such as identifiable cancer, pregnancy, or major surgery. In contrast, unprovoked VTE is considered to occur without any identifiable risk factors.⁴

It is worth noting the close relationship between VTE and cancer burden. On one hand, patients with cancer are at significantly higher risk of VTE owing to abnormal coagulation functions, bed rest, and immobilization, and so on, compared with populations without cancer.⁵ On the other hand, VTE (especially unprovoked VTE) may be the first manifestation of occult cancer.6, 7, 8 Studies show that the risk of occult cancer significantly increases in newly diagnosed VTE patients within 6 to 12 months after the thrombotic event. Within 1 year of an unprovoked VTE diagnosis, 6% to 15% of patients are found to have cancer.^9,10 Additionally, VTE and its related complications further exacerbate the disease burden of patients with cancer and are associated with advanced cancer stages and poor prognosis.^11,12

Clinically, screening for occult malignancies in VTE patients is crucial for early cancer detection and intervention to decrease related morbidity and mortality.⁶ Over the past decade, the investigation of screening methods that offer the best diagnostic rates for the identification of occult malignancies in this patient population has been the focus of numerous studies.¹³ This review comprehensively organizes the epidemiological characteristics, risk stratification, and screening strategies related to unprovoked VTE and occult cancer, aiming to provide scientific guidance for clinical practice and reference for improving patient outcomes.

Epidemiology

Since Trousseau first revealed the subtle relationship between cancer and VTE in 1865, more and more scholars have focused on it. In the 1990s, an epidemiological study by Nordström et al¹⁴ followed 3795 DVT patients without a cancer diagnosis and found that the incidence of cancer within 6 months after DVT diagnosis was five times that of the general population. In the same decade, a national cohort study by Sørensen et al¹⁵ investigated 15,348 DVT patients and 11,305 PE patients, discovering that the standardized incidence ratio (SIR) for cancer within 6 months after DVT and PE was 1.3. The SIR is used to measure the ratio of the incidence of a certain disease in a specific study population to the incidence rate in a standard reference population (eg, a general population with similar age and sex composition). A systematic review in 2008, which integrated 34 original studies totaling 9516 individuals, showed that 10% of unprovoked VTE patients were diagnosed with cancer in the first year of follow-up.¹⁰ However, a comprehensive report published in 2017, which included 22 epidemiological studies of East Asian populations, indicated that 7% of VTE patients were also diagnosed with cancer within 2 years of a VTE diagnosis.¹⁶ Another meta-analysis in 2020, which included 10 prospective studies and 2381 patients, found a lower diagnosis rate of covert cancer within 1 year in unprovoked VTE patients, at 5.4%.¹⁷ Concurrently, in the latest original studies from the past 3 years, the cancer diagnosis rate during follow-up for unprovoked VTE patients ranged from 2.4% to 8.8%, which is lower than 10.0% (Table I).18, 19, 20, 21, 22, 23, 24, 25 The decline in the diagnosis rate of occult cancer in patients with unprovoked VTE may be related to medical screening technology advancements (eg, more accurate imaging and biomarker tests) and upgrades in clinical management protocols. Unprovoked VTE, such as PE or DVT, is an independent risk factor for occult malignancies, meaning that they have a greater chance of having cancer than the average person.

Table I.

Summary of cancer incidence rate in recent research

Study ID	Study type	Sample size	Incidence rate of cancer
			Six months	First year	Other times
Mulder et al (2023)18	Prospective multicenter cohort study	476	–	5.3% (25/476)	–
Rosell et al (2023)19	Prospective multicenter cohort study	417	–	7.0% (29/417)	–
Cordeanu et al (2022)20	Prospective cohort study	993	–	5.3% (53/993)	–
Sánchez-López et al (2024)21	Nested case-control study	214	–	8.8% (19/214)	–
Franco-Moreno et al (2024)22	Multicenter retrospective cohort study	(1) RIETE group: 815 (2) SOME group: 418	–	(2) 7.9% (33/418)	(1) Two years 6.9% (56/815)
Yamashita et al (2023)23	Retrospective multicenter cohort study	3706	–	3.7% (135/3706)	Three years 7.0% (259/3706)
Pandit et al (2022)24	Retrospective cohort study	97,754	2.4% (2354/97754)	–	–
Lee et al (2023)25	Retrospective cohort study	150	–	–	Two years 6% (9/150)

RIETE, Computerized registry of thromboembolic disease; SOME, screening for occult malignancy in patients with idiopathic venous thromboembolism.

Pathophysiological mechanism

Malignant cancers induce thrombosis

Malignant tumors form complex interactions with host hemostatic systems through multidimensional mechanisms that drive hypercoagulable states and thrombotic events. The core pathophysiological processes involve abnormal stimulation of the cascade of coagulation, dysregulation of the inflammatory response, suppression of fibrinolysis, and tumor-platelet interactions, which together constitute a dynamically evolving prothrombotic network.²⁶ Tumor cells directly release procoagulant substances (eg, cancer procoagulant factor and tissue factor [TF]) and TF-rich particles, triggering fibrin deposition via exogenous coagulation pathways.²⁷ Experimental studies have shown that injecting mice with TF-positive particles induced a disseminated intravascular coagulation-like syndrome, suggesting a key role for circulating particles in systemic thrombosis.²⁸ Notably, TF expression levels are positively correlated with tumor aggressiveness, and it has been found that, in pancreatic cancer, sustained high expression of TF activates coagulation, triggers venous thrombosis, and may also trigger distal embolic events through particulate spread.^29,30

Proinflammatory elements in the tumor microenvironment, including tumor necrosis factor-α, interleukin-1β (IL-1β), and IL-6, form a prothrombotic microenvironment by activating adhesion molecules of endothelial cells (such as vascular cell adhesion molecule-1 and intercellular adhesion molecule-1) and inducing leukocyte and platelet adhesion.³¹ At the same time, procoagulant enzymes are released and surface TF is revealed when monocytes/macrophages and neutrophils are activated, further amplifying the coagulation signal. In patients with cancer, activated platelets cannot only directly trigger a coagulation cascade reaction and promote thrombosis, but also go through complex interactions with other types of blood cells (eg, leukocytes). It has been discovered that leukocytes release neutrophil extracellular traps (NETs) in response to activated platelets and that the interaction between endothelial cells and these structures and platelet-mediated processes is closely connected to the development of cancer-associated VTE.²⁶ Meanwhile, the hypercoagulable state is further exacerbated by the aberrant interaction between tumor cells and platelets. Cancer cells activate platelets by secreting adenosine diphosphate, thrombin, and matrix metalloproteinases,³² whereas P-selectin-mediated tumor-platelet aggregates on the surface of platelets not only promote thrombosis, but also tumor metastasis by shielding immune recognition.³³

The fibrinolytic system's suppression is a key mechanism for maintaining the stability of pathological thrombus. Tumor cells inhibit tissue-type plasminogen activator activity by upregulating plasminogen activator inhibitor-1, causing impaired fibrin degradation. Clinical observations have demonstrated a substantial correlation between the prevalence of VTE and high plasminogen activator inhibitor-1 levels in patients with cancer.^34,35

VTE may promote cancer progression

Inflammatory responses are present throughout the pathogenesis of VTE and are associated with cancer development. Inflammatory factors such as tumor necrosis factor-α, IL-6, and IL-8 are present at higher concentrations in VTE patients, and these factors not only promote the activation of vascular endothelial cells, but also affect the adhesion and aggregation of blood cells, which creates favorable conditions for the adhesion, invasion, and metastasis of tumor cells.³⁶ For example, IL-6 promotes TF production and factor VIII transcription, and it also promotes the synthesis of fibrinogen. In the meantime, TF is the primary promoter of the exogenous coagulation pathway, and tumor cell metastasis and hypercoagulability are strongly correlated with its aberrant expression.^37,38

In addition, an observational study by Rosell et al¹⁹ discovered that VTE patients with concomitant occult cancers had considerably higher circulating plasma levels of H3Cit-DNA, a hallmark of NETs, than non-VTE patients, indicating a link between circulating H3Cit-DNA and occult cancers. Given the critical role of NETs in inducing tumor cell progression, NETs might serve as a bidirectional bridge that both participates in tumor microenvironment-driven VTE formation and facilitates the developmental process of occult cancers in VTE patients.

Risk stratification

The population characteristics of unprovoked VTE

Age

The Screening for Occult Malignancy in Patients with Idiopathic Venous Thromboembolism (SOME) score appeared in the SOME trial's post hoc analysis in Canada. The Registro Informatizado Enfermedad TromboEmbólica (RIETE) score was created from a global multicenter prospective cohort based on RIETE. Advanced age constituted a separate risk factor in both predictive models. The SOME trial³⁹ was a randomized controlled trial that included 854 first-time unprovoked VTE patients from 2008 to 2014, randomly divided into a limited screening group (n = 431) and a limited screening + computed tomography (CT) group (n = 423), to explore the clinical significance of extensive screening over a 1-year follow-up. Subsequent researchers conducted a secondary analysis and found that the hazard ratio (HR) for cancer risk in unprovoked VTE patients aged ≥60 years was 3.11 (95% confidence interval [CI], 1.41-6.89; P = .005), with a 3.9% incidence of occult cancer (95% CI, 2.8%-5.4%).⁴⁰ The RIETE study was an observational study that included 5863 acute unprovoked VTE patients from 2001 to 2014, with a median follow-up time of 2 years. A comparison between the case group (n = 444) and the control group (n = 5419) found that the HR for cancer risk in unprovoked VTE patients aged ≥70 years was 1.90 (95% CI, 1.55-2.33, P < .001), with a high incidence of occult cancer at 12.1% (95% CI, 10.3%-14.1%).⁴¹ In addition, an individual patient data meta-analysis integrated 2316 unprovoked VTE patient data from 10 prospective studies, showing that the 1-year cancer prevalence in patients aged ≥50 years was seven times that of patients <50 years old, and the 1-year cancer prevalence in patients aged ≥80 years reached 9.1% (95% CI, 5.6%-15%).⁴² Another meta-analysis including 4 prospective studies explored 332 patients with first occurrence of unprovoked VTE and showed that among patients aged ≥50 years, the incidence of occult cancer was higher than those <50 years of age (6.6% vs 1.0%).⁴³ The detailed data are presented in the Supplementary Table (online only). It is possible that the age boundary values are not uniform owing to differences in the characteristics of the enrolled subjects and the sample size. However, undoubtedly, older unprovoked VTE patients have a significantly higher risk of detecting cancer than younger patients.

Sex

A meta-analysis that included 10 original studies demonstrated that 5.7% of men with VTE and 5.0% of women with VTE were diagnosed with cancer within 1 year.⁶ A multicenter, randomized, controlled, prospective trial (MVTEP study) by Robin et al⁴⁴ showed a significantly higher occurrence of cancer in male patients than in females at 2 years (8.7% vs 3.8%; P = .04). Paradoxically, the first national retrospective cohort study by Pandit et al²⁴ recently demonstrated that being female was a risk factor for unprovoked VTE-associated occult cancer (Supplementary Table, online only). Therefore, it is necessary to investigate in larger prospective studies whether sex is an independent risk indicator of occult malignancy in individuals with unprovoked VTE.

Pathological conditions

Christensen et al⁴⁵ carried out a population-based nationwide cohort study to assess VTE and cancer risk in patients with diabetes. Among the 8783 patients with diabetes included with a first diagnosis of VTE, 4.1% of patients were diagnosed with cancer within 1 year, and the SIR was 3.28. Among these, patients <75 years and female patients were more likely to manifest cancer. This finding suggests that diabetes may increase cancer risk through other mechanisms (eg, systemic inflammation, coagulation abnormalities). In 2021, a study by Thomsen et al,⁴⁶ drawn from the Danish National Health Registry's data, assessed the malignancy risk in patients with VTE combined with diverticular disease. Among the 3406 patients with concurrent diagnoses of diverticular disease and VTE, the risk of cancer in the first year was significantly elevated, with an incidence rate of 6.2% and 2.9 (SIR). In the same year, Fuglsang et al⁴⁷ revealed that patients with dementia who also had concomitant VTE had an increased chance of developing cancer. Among the 3552 patients, the first-year cancer incidence rate was 2.8% and 1.9 (SIR). In another study, 11,759 patients with VTE who had used aspirin were divided into three groups according to the duration of use, with malignancy incidence rates of 6.0% to 6.7% (no statistical difference in incidence rates among the three groups) and SIRs of 2.8 to 3.3 after 1 year of monitoring.⁴⁸ In summary, patients with VTE who have diabetes, diverticular disease, dementia, or have taken aspirin may have a higher risk of cancer occurrence compared with the general population (Fig 1). Although the number of studies focusing on this association is limited, the included studies have large sample sizes, which, to some extent, ensures the reliability of the current conclusion; further high-quality studies are still needed to verify the stability and generalizability of this conclusion.

Fig 1.

Incidence and standardized incidence ratio (SIR) of occult cancers within 1 year in venous thromboembolism (VTE) patients with comorbid other pathologies. The middle circle refers to a possible bidirectional relationship between cancer and VTE. (A1) People with diabetes comorbid with VTE. (A2) People with diverticular disease comorbid with VTE. (A3) People with dementia comorbid with VTE. (A4) People who have taken aspirin for the VTE population. IL, interleukin; NET, neutrophil extracellular trap; TNF, tumor necrosis factor.

The locations and types of unprovoked VTE

The site and type of VTE may be influential factors for occult cancer. As early as the 1990s, in contrast with lower extremity DVT, upper extremity DVT (U-DVT) was found to be more strongly related to occult cancer.⁴⁹ Adelborg et al⁵⁰ conducted a cohort study to assess subsequent cancer risk in patients with U-DVT. The results showed that, in 1087 patients with U-DVT, the SIR in the first 6 months was 12.58 (95% CI, 9.58-16.23).⁵⁰ Another study demonstrated that the SIR of cancer in patients with superficial vein thrombosis, DVT, and PE was 2.46 (95% CI, 2.10-2.86), 2.75 (95% CI, 2.60-2.90), and 3.27 (95% CI, 3.03-3.52), respectively, and that, compared with the general population, people with PE were more likely to develop cancer.⁵¹ Interestingly, the one-year cancer incidence rates of patients in the DVT, PE, and DVT + PE groups (5.6%, 4.3%, and 5.6% separately) (Supplementary Table, online only) did not differ significantly, according to a meta-analysis that included 10 trials.⁵²

Some specific sites of VTE, such as splanchnic venous thrombosis (SVT), retinal vein thrombosis, cerebral venous thrombosis (CVT), or even arterial thrombosis, may be indicative of occult cancer. Indeed, in a study, 183 cases of cancer were identified in 1191 SVT patients over a median follow-up of 1.6 years, with a risk of cancer in the first 3 months of 8.0%, SIR of 33 (95% CI, 27-40), and an increased risk of hepatocellular carcinoma and pancreatic carcinoma (SIRs of 1805 and 256).⁵³ Recently, the Dutch scholar Munckhof⁵⁴ assessed the likelihood of a new cancer diagnosis in patients after CVT. Of the 2649 CVT patients included, 119 (4.5%) were diagnosed with cancer during a median follow-up of 4.7 years (range, 1.9-8.9 years), their SIR was 3.35 (95% CI, 2.41-4.55) within the first year, and their risk of developing cancer was higher in males and those ages ≥50 years (P < .01) (Supplementary Table, online only). According to a study, people with lower limb arterial thrombosis have a 6-month cancer risk that is more than three times higher than the general population.⁵⁵ Nevertheless, retinal vein thrombosis does not appear to increase the incidence of occult cancer⁵⁶ (Fig 2).

Fig 2.

Occult cancer incidence and standardized incidence ratio (SIR) for patients with venous thromboembolism (VTE) at specific sites over the corresponding period. (B1) Population with retinal vein thrombosis. (B2) Population with cerebral venous thrombosis (CVT). (B3) Population with splanchnic venous thrombosis. (B4) Population with lower extremity arterial thrombosis.

Correlation between recurrent VTE and occult cancer

In a prospective cohort study⁵⁷ investigating the risk of occult cancer among patients with recurrent VTE, 197 patients with two unprovoked VTE episodes occurring within a 2-year interval were enrolled. The incidence of cancer within 1 year after the second VTE was 9.19%. Notably, patients with an interval of <1year between the first and recurrent VTE had a substantially higher 1-year cancer HR of 18.54 (95% CI, 2.02-169.81) (Supplementary Table, online only), indicating that a short interepisode duration of <1 year is an independent risk factor for a subsequent cancer diagnosis. However, owing to the limited sample size of this study, these findings should be interpreted with caution, and further research with larger cohorts is warranted to clarify the association between recurrent VTE and occult malignancy.

In another study,⁵⁸ the 1-year VTE recurrence rate was reported to be 38.6% (95% CI, 27.8%-52.3%) in patients with occult cancer, 15.5% (95% CI, 10.2%-23.3%) in those with overt cancer, and only 3.8% (95% CI, 2.5%-5.6%) in cancer-free individuals. When compared with patients without cancer, the risk of recurrence was 12-fold higher in the occult cancer group (HR, 12.4) and four-fold higher in those with overt cancer (HR, 4.3), suggesting that individuals who have occult cancer are more likely to have a recurrent VTE (Supplementary Table, online only).

Screening strategies

Unprovoked VTE patients require cancer screening

Unprovoked VTE often serves as the first manifestation of occult cancer, and more than one-half of patients with unprovoked VTE who undergo screening for occult cancer have already reached advanced stages or developed metastasis, making tumor screening for unprovoked VTE patients highly necessary.^11,20,58, 59, 60 Meanwhile, they exhibited a higher all-cause mortality from all causes in 3 years, with cancer accounting for the leading cause of death (48.6%).⁶¹ Furthermore, one study reported that 88% of patients who died of VTE complicated by occult cancer succumbed to cancer-related causes.⁶² Another study indicated that, in the VTE population with newly diagnosed cancer, 76.3% of patients died from cancer.²³ These data invariably reveal that patients with unprovoked VTE may have serious consequences if they have concurrent cancer, which perhaps emphasizes the importance and necessity of early identification and diagnosis of cancer.

The selection of screening strategies

The International Society on Thrombosis and Hemostasis states in its latest guidelines⁶³ that patients with unprovoked VTE should undergo either limited or extensive cancer screening. Limited screening covers history, blood tests (lactate dehydrogenase, electrolytes, creatinine, liver function tests, and complete and categorical blood counts), physical examination, radiographs of the chest, and cancer screening according to age and sex. Extensive screening adds one or more additional CT scans of the chest, abdomen, and pelvis, endoscopy, mammography, abdominal and pelvic ultrasound, serum tumor markers, and 18F-fluorodesoxyglucose positron emission tomography with CT (FDG PET/CT).

Numerous investigations have been carried out to identify the prevalent forms of concurrent occult cancers in VTE patients. A meta-analysis¹⁶ reported that lung, liver, and colorectal cancers were the most common, accounting for 18.3%, 12.3%, and 10.9% of cases, respectively. Another meta-analysis⁵² identified colon cancer as the most prevalent type, constituting 16% of all detected cancers, followed closely by lung (14%), pancreatic (11%), and hematological cancers (11%). Regarding gender differences, one study⁶⁴ revealed that the most common cancers in men are prostate and lung cancers, with the risk of lung cancer being seven times higher than in women. In contrast, cancer types in female patients are more diverse, with a relatively high incidence of colorectal and breast cancers. Several recent studies^20,23,65 have further corroborated that occult cancers are frequently concentrated in sites such as the lung, colorectum, prostate, breast, and hematological system. These findings have important implications for clinical practice, because they may assist physicians in conducting more targeted cancer screening in patients with unprovoked VTE, thereby improving early diagnosis and treatment outcomes.

Some studies have explored screening strategies. A multicenter, randomized controlled trial (RCT) by Carrier et al³⁹ evaluated the efficacy of two screening strategies. The trial enrolled 854 patients with a first episode of unprovoked VTE, who were randomized to either limited screening (n = 431) or limited screening plus CT (n = 423). Comparison between the two groups showed that the cancer detection rate, missed diagnosis rate, and overall mortality rate did not differ significantly during the 1-year follow-up (P = .28, P = 1.0, and P = .75, respectively). Another RCT⁶⁶ similarly reported that the absolute difference in the cancer detection rate within 1 month was only 2% between the limited screening group and the limited screening plus CT group, a difference that was not statistically significant (P = .81). In addition, a meta-analysis⁶⁷ published in 2017 included five controlled studies. The extensive screening strategies used in these studies are based on limited screening and incorporate additional imaging examination methods. Among them, four studies used thoracoabdominal pelvic CT scans; only one study used whole-body ¹⁸F-FDG PET/CT. This result demonstrated that the risk of a missed diagnosis and cancer-related death did not differ significantly between the intensive screening group and the limited screening group (P sequentially = .15 and .65).

Notably, in 2004, Piccioli et al⁶⁸ published a prospective randomized clinical trial, where the extensive screening protocol included abdominal (including pelvic) ultrasound, thoracoabdominal pelvic CT scans, gastroscopy or double contrast barium meal, colonoscopy or sigmoidoscopy combined with barium enema, fecal occult blood test, sputum cytology examination, tumor marker detection, and for women, mammography + Pap smear, and for men, transabdominal prostate ultrasound + prostate-specific antigen test. The study results showed that, compared with the control group, the average cancer diagnosis delay time in the extended screening group was significantly shorter (1.0 month vs 11.6 months; P < .001), and the proportion of early tumors detected was higher (64.3% vs 20%; P = .047); however, there were no statistically significant differences in the primary outcome (cancer-related mortality: 2.0% vs 3.9%; P > .05) and the secondary outcome (cancer-related death and residual/recurrent malignant tumors: 5.1% vs 7.9%; P > .05). It should be noted that the study sample size was only 201 cases, and the generalizability of its conclusions still needs to be validated by more studies. Interestingly, the use of FDG PET/CT may be beneficial for the precise identification of occult cancers. A multicenter RCT by Robin et al⁶⁹ demonstrated that VTE patients with negative initial FDG PET/CT screening were significantly less likely to be diagnosed with a subsequent cancer than those with negative limited screening (1/186 vs 9/193; P = .02). Another of Robin et al’s studies⁷⁰ reported that FDG PET/CT exhibited high accuracy in detecting occult cancer, with a specificity of 85% and sensitivity of 90%. In an RCT⁷¹ evaluating the benefits of screening, although the overall detection rate did not differ significantly between the extensive screening group and the limited screening group that included FDG PET/CT (6.7% vs 6.1%; P = .76), the former demonstrated a considerably higher initial screening detection rate than the latter (5.6% vs 2.0%). Another meta-analysis⁴³ further assessing the accuracy of FDG PET/CT confirmed its high diagnostic performance in identifying cancer in patients with unprovoked VTE, with a sensitivity of 87.3%, specificity of 70.2%, positive predictive value of 17.9%, and negative predictive value of 98.9%.

In general, combining the guidelines⁶³ and the latest research evidence, we recommend routine limited screening plus age-/sex-specific screening for patients with unprovoked VTE for the first time. Recurrent (especially when the interval between the first and recurrent VTE is <1year) VTE, special locations of thrombosis (such as CVT, splanchnic venous thrombosis, lower extremity arterial thrombosis) and other pathological state populations (such as patients with diabetes, diverticulum disease, dementia, and aspirin use history) can also undergo routine cancer screening. For those suspected of having tumors but with unclear other examinations, extensive screening methods such as FDG PET/CT can be considered for early diagnosis.

Cancer prediction scoring models

Since the RIETE score⁴¹ and SOME score⁴⁰ were proposed, research on their prognostic value in occult cancer in patients with unprovoked VTE has expanded considerably. Indeed, an examination of the Hokusai-VTE study after the fact (a multinational, randomized, double-blind clinical trial; n = 8240)⁷² showed the RIETE score's area under the receiver operating characteristic curve (AUC) was 0.62. and the AUC of 0.59 for the SOME score, with both scores showing relatively limited performance. In another meta-analysis,¹⁷ although the high-risk group's cancer risk (defined as ≥3 points) after dichotomous treatment of the RIETE score was twice as high as that in the low-risk group (defined as ≤2 points) (HR, 2.0; 95% CI, 1.3-3.4), its AUC was 0.59 (95% CI, 0.52-0.66), which had a poorer predictive power. In contrast, the AUC for the SOME score was 0.56. There was no significant difference in cancer risk between the high-risk group and the low-risk group after dichotomization (HR, 1.2; 95% CI, 0.55-2.7). In 2024, the results of Franco-Moreno et al²² also demonstrated the low effectiveness of the RIETE and SOME scores (AUCs of 0.430 and 0.351 in that order). In summary, the two current scoring models do not meet clinical needs, and new predictive models or prospective studies must be created to further validate the feasibility of existing models.

In 2022, a scoring model, the VTE Malignancy Score, was proposed in a national study²⁴ with the following components: Total score = Female (1.08 points) + Age >65 (1.06 points) + Inferior vena cava Charlson comorbidity index ≥5 (1.38 points) + Upper extremity thrombosis (1.37 points) + Inferior vena cava thrombosis (2.24 points), with a score of ≥3 indicating high risk. This predictive model had an internally validated AUC of 0.74, with 86% sensitivity and 89% specificity when the threshold was ≥3 points, a good internal effect, but more external studies are needed to demonstrate its efficacy. A recent study⁶⁵ included 815 cases of unprovoked VTE patients from 2005 to 2021, of which 56 cases (accounting for 6.9%) were diagnosed with occult cancer during the follow-up period (30 days to 24 months after thrombotic events), forming the cancer subgroup in this study. The study developed a cancer risk prediction model for unprovoked VTE patients using machine learning (ML) technologies, with models trained using the XGBoost, LightGBM, and CatBoost algorithms. According to the findings, the CatBoost model was the most predictive with a positive predictive value of 75%, negative predictive value of 93%, and AUC of 0.86 with key variables including age, sex, weight, heart rate, systolic blood pressure, previous VTE, D-dimer, alanine aminotransferase, hemoglobin, serum creatinine, cholesterol, platelets, triglycerides, leukocyte count, and chronic lung disease. The study revealed for the first time the potential value of ML in patients with unprovoked VTE, occult cancers can be predicted, which lays the foundation for the development of subsequent high-performance prediction models. However, the modest sample size of the study's cancer subgroup (6.9%) may limit the generalizability of the model's performance, so more large-sample studies are needed in the future to further validate the application value of ML in this field (Table II).

Table II.

Summary of prediction models and predictive efficiency

Study ID	Prediction models	Predictors	AUC	Sensitivity, %	Specificity, %	PPV, %	NPV, %
Jara-Palomares et al (2016)41	RIETE model	Male (+1), age ≥70 years (+2), chronic lung disease (+1), anemia (+2), PLT ≥350∗109/L (+1), recent surgeries (−2), previous VTE (−1)	0.64 (95% CI, 0.61-0.66)	–	–	–	–
Kraaijpoel et al (2018)72			0.62 (95% CI, 0.57-0.66)	–	–	–	–
Mulder et al (2020)17			0.59 (95% CI, 0.52-0.66)	–	–	–	–
Franco-Moreno et al (2024)22			0.43 (95% CI, 0.38-0.47)	32	76	9	94
Kraaijpoel et al (2018)72	SOME model	Age ≥60 (+1), current smoking (+1), previous VTE (+1)	0.59 (95% CI, 0.55-0.62)	–	–	–	–
Mulder et al (2020)17			0.56 (95% CI, 0.46-0.65)	–	–	–	–
Franco-Moreno et al (2024)22			0.35 (95% CI, 0.27-0.43)	33	88	20	94
Pandit et al (2022)24	VTE Malignancy Score	Age >65 (+1.06), female sex (+1.08), IVC thrombosis (+2.24), upper extremity thrombosis (+1.37), CCI ≥5 (+1.38)	0.74 (95% CI, 0.73–0.75)	86	89	–	–
Franco-Moreno et al (2025)65	CLOVER Score	Age, sex, systolic blood pressure, heart rate, weight, chronic lung disease, D-dimer, alanine aminotransferase, hemoglobin, serum creatinine, cholesterol, platelets, triglycerides, leukocyte count, previous VTE	0.86 (95% CI, 0.83-0.87)	62	94	75	93

AUC, Area under the receiver-operating characteristics curve; CCI, Charlson Comorbidity Index; CI, confidence interval; CLOVER, development of a predictive model of occult cancer after venous thromboembolism event using machine learning; IVC, inferior vena cava; NPV, negative predictive value; PLT, platelet; PPV, positive predictive value; VTE, venous thromboembolism.

It is worth noting that a recent methodological study⁷³ describes the design protocols of two ongoing prospective studies: the SOME-RIETE trial and the ValRIETE study. Among them, the ValRIETE study is a prospective cohort study conducted across an international multicenter setting, planning to enroll 1550 patients with first-time unprovoked VTE, with 1-year and 2-year follow-up periods. This study is expected to enable the external validation of the RIETE score. In contrast, 650 high-risk individuals with an RIETE score of ≥3 are anticipated to be enrolled in the SOME-RIETE trial, an RCT that compares restricted screening with combined FDG PET/CT screening. The results of these two studies are worth looking forward to.

Biomarkers for cancer screening

A biomarker represents a measurable and assessable feature that serves to identify physiological functions, disease progression, or the effects of a medicinal treatment. These features encompass multiple levels, including molecular and cellular, and hold various applications in the early detection and monitoring of diseases.⁷⁴ In the cancer management system, the application of biomarkers spans the entire process, playing a pivotal role especially in early detection, risk stratification, and screening.⁷⁵ According to a cohort study,¹¹ patients with VTE and a D-dimer level of >4000 ng/mL had a 4.12-fold higher incidence of occult cancer compared with those with a level of <2000 ng/mL (HR, 4.12; P < .01). The results of another prospective study²¹ presented that soluble P-selectin levels in the cancer group (median, 72 ng/mL) were approximately twice those in the non-cancer group (median, 41 ng/mL; P < .05). The specificity for diagnosing occult cancer reached 91% (95% CI, 79.9%-96.7%) in the presence of both a D-dimer level of >10,000 μg/L and a soluble P-selectin level of >62 ng/mL. More innovatively, a prospective cohort study¹⁹ reported that, for every 500 ng/mL increase in H3Cit-DNA levels among VTE patients, the risk of occult cancer increased by 79%. These findings suggest that H3Cit-DNA may serve as a risk factor for cancer in VTE patients and demonstrate potential predictive value. Early studies demonstrated that platelet RNA sequencing can accurately distinguish patients with malignancy from healthy individuals.⁷⁶ Paradoxically, a multicenter prospective cohort research by Mulder et al¹⁸ reported that the accuracy of platelet RNA sequencing for cancer diagnosis was relatively low (AUC, 0.54; 95% CI, 0.41-0.66). It should be noted, however, that the sample size in this study was relatively small, and further investigations with larger cohorts are needed to evaluate the clinical utility of platelet RNA sequencing. Unfortunately, tumor markers, which are recognized as tumor-specific substances, have been studied relatively little. An analysis conducted by Chinese scholars Yang et al⁷⁷ noted that among the six tumor markers (carcinoembryonic antigen [CEA], alfa fetoprotein, CA125, CA199, CA242, and CA724), the AUC of CEA (0.716; 95% CI, 0.654-0.777), CA199 (0.732; 95% CI, 0.647-0.791) were greater in diagnostic efficacy. In another study,⁶⁰ a statistically significant variation was seen in the neutrophil-lymphocyte ratio between cancer and noncancer groups (4.4 and 3.0, respectively; P < .05), and the sensitivity of primary cancer screening was 80% when the neutrophil-lymphocyte ratio was combined with three tumor markers, namely, CEA, CA199, and CA125. Although tumor markers are commonly used for cancer surveillance, based on these results, it is speculated that they may serve as auxiliary indicators for early screening of occult cancer in unprovoked VTE patients. Further studies are required to validate the screening value of tumor markers.

Conclusions

Unprovoked VTE events and tumor lesions are both common clinical conditions with bidirectional associations. Patients with tumors are prone to concurrent VTE events in multiple body sites, the latter of which can increase mortality in patients with tumors. Patients with unprovoked VTE have a greater risk of being diagnosed with occult cancer than the general population. Routine limited screening for unprovoked VTE patients, especially elderly patients, special location VTE, VTE under certain pathological conditions, and recurrent VTE patients, may be beneficial for early diagnosis and treatment of cancer in patients, improving patient prognosis. However, the application value of extensive screening needs more high-quality research to prove. Several predictive models are currently available for the clinical prediction of occult cancers; however, their validity still requires further verification. It is anticipated that with ongoing research, more effective methods will be developed to identify occult cancer in patients with VTE.

Author contributions

Conception and design: JW, QZ

Analysis and interpretation: JW, QZ

Data collection: JW, CZ, FW

Writing the article: JW, CZ

Critical revision of the article: FW, QZ

Final approval of the article: JW, CZ, FW, QZ

Statistical analysis: Not applicable

Obtained funding: Not applicable

Overall responsibility: QZ

JW and CZ contributed equally to this article and share co-first authorship.

Funding

None.

Disclosures

None.