Abstract
Background
The nonspecific nature of pulmonary embolism (PE) in patients initially diagnosed as symptomatic deep vein thrombosis (DVT) often leads to instances of underdiagnosis. The thrombus density of the DVT could serve as a predictive factor for PE. Dual-energy (DE) computed tomography (CT) obtains extra attenuation measurements to differentiate thrombus composition. Therefore, this retrospective study aimed to investigate whether DE-CT venography (DE-CTV) parameters of the DVT could predict the asymptomatic PE occurrence in a single tertiary institution in China.
Methods
Between March 2020 and October 2023, patients with DVT diagnosed through DE-CTV were retrospectively identified. Thrombus density ratio, normalized iodine concentration (NIC, %), normalized effective atomic number (NZeff, %), and the slope of the spectral Hounsfield unit curve (λHU) were calculated. Risk factors associated with PE were identified using multivariate logistic regression analysis. Receiver operating characteristic (ROC) analysis was used to assess the value of these parameters for predicting the risk of PE.
Results
Finally, a total of 49 patients were included in this study, including 24 patients in the DVT-only group and 25 patients in the DVT-PE group. Thrombus density ratio, NIC, NZeff, and λHU were significantly higher in the DVT-PE group than in the DVT-only group (P<0.05). Multivariate analyses revealed that NIC was an independent predictor of PE [odds ratio =1.25, 95% confidence interval (CI): 1.09–1.53; P=0.009]. ROC analysis showed that NIC >24.4% was the optimal cutoff for predicting PE with an area under the curve of 0.920 (95% CI: 0.846–0.994, P<0.001).
Conclusions
The NIC derived from DE-CTV is independently associated with asymptomatic synchronous PE in patients with lower-extremity DVT. The proposed NIC cutoff value is exploratory and requires further validation.
Keywords: Pulmonary embolism (PE), computed tomography (CT), dual-energy (DE), venography
Highlight box.
Key findings
• Dual-energy computed tomography venography (DE-CTV) parameters—specifically normalized iodine concentration (NIC) and thrombus density ratio—are significantly higher in deep vein thrombosis (DVT) patients with concurrent, asymptomatic pulmonary embolism (PE) compared to those without PE. Intra- and inter-observer reliability for these measurements demonstrated acceptable to good reproducibility.
What is known and what is new?
• The subtle and nonspecific clinical presentation of PE frequently results in its underdiagnosis among patients with symptomatic DVT. Consequently, assessing DVT thrombus density may serve as a potential predictive indicator for the presence of concurrent PE.
• This study explores a novel application of DE-CTV, suggesting that quantitative spectral parameters of the lower extremity DVT itself might serve as predictive biomarkers for concurrent PE.
What is the implication, and what should change now?
• For symptomatic DVT patients undergoing DE-CTV, elevated values in thrombus density ratio or NIC could be considered as indicators raising clinical suspicion for concurrent PE.
• This quantitative approach might help optimize clinical workflows by aiding in the triage of DVT patients who could benefit from prompt CTPA screening or early evaluation for anticoagulant therapy. Future prospective validation of these DE-CTV parameters is warranted to confirm their clinical utility.
Introduction
Venous thromboembolism (VTE), which includes deep vein thrombosis (DVT) and pulmonary embolism (PE), is the third most common cardiovascular disease (1). It is estimated that about 10 million people develop VTE each year worldwide; moreover, the incidence of VTE increases with age (1,2). As a complication of DVT, PE can be life-threatening. Therefore, early diagnosis and treatment of PE are important to improve the prognosis of these patients. Despite the existence of various clinical models for diagnosis, the nonspecific nature of PE symptoms often complicates accurate detection, leading to instances of underdiagnosis (3,4). While computed tomography pulmonary angiography (CTPA) is effective in diagnosing PE with high sensitivity and can be used for risk stratification, concerns persist regarding its overutilization and consequent radiation exposure risks (3,5,6).
Timely diagnosis and management of DVT are needed as they constitute the primary cause of PE. Currently, venous ultrasound remains the primary imaging modality for diagnosing DVT (7). CT venography (CTV) can provide high-resolution images and is equally effective for diagnosing DVT, offering detailed information about deep veins such as the iliac vein and inferior vena cava (7,8). It may be suitable for patients suspected of having PE, experiencing venous compression, or requiring endovascular treatment (7,9). The thrombus density on CT is correlated with the proportion of its components, such as red blood cells and fibrin (10,11). Thus, the thrombus density may provide indications of thrombus age and composition, and previous studies have found that the thrombus density in DVT may serve as a predictive factor for PE (12-14).
Dual-energy (DE) CT obtains extra attenuation measurements using a second X-ray spectrum, enabling the distinction of various materials or the quantification of mass density in mixtures with known composition (15). DE CT used for vascular imaging can enhance vascular contrast, reduce contrast agent and radiation exposure, and generate material decomposition images for distinguishing different materials with high attenuation values (16). Recent study further emphasizes the growing role of advanced spectral CT in quantifying thrombus iodine uptake to differentiate acuity and composition (17). However, little is known about the clinical value of DE CT-derived parameters in patients with DVT. Therefore, the purpose of our study is to investigate whether DE-CTV parameters in patients with DVT are associated with the occurrence of PE. We present this article in accordance with the STROBE reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1-0250/rc).
Methods
The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments, and was approved by the Ethics Committee of The First Affiliated Hospital with Nanjing Medical University (protocol code: 2024-SR-517). As this was a retrospective study, informed consent was waived. Data utilized in this study can be accessed upon reasonable request for collaboration, secondary analysis, or additional investigation.
Study participants
In The First Affiliated Hospital with Nanjing Medical University, DE-CTV is primarily indicated for patients with extensive or severe symptomatic DVT who are active candidates for endovascular treatments (e.g., catheter-directed thrombolysis or mechanical thrombectomy). This clinical pathway naturally selects for a cohort with a higher burden of iliofemoral thrombosis. Then, all patients diagnosed with lower-extremity DVT undergo a systematic clinical evaluation, which includes an assessment of symptoms, D-dimer levels, and predisposing risk factors. Patients identified as having a high risk for PE based on this comprehensive assessment routinely undergo concurrent CTPA to guide subsequent treatment strategies, regardless of the presence of overt PE symptoms. This study specifically targeted the subset of patients who lacked dominant PE symptoms (such as acute chest pain, severe dyspnea, or hemoptysis) but underwent CTPA due to their high-risk profile.
From March 2020 to October 2023, we reviewed the medical records of DVT patients without overt PE features undergoing concurrent CTPA in The First Affiliated Hospital with Nanjing Medical University. Inclusion criteria included: (I) patients aged 18 to 80 years; (II) patients diagnosed with symptomatic DVT (involving common/superficial femoral vein, common/external iliac vein) by DE-CTV; (III) patients with CTPA examination synchronously. Exclusion criteria included: (I) patients with iliac vein compression; (II) patients with VTE history; (III) patients with malignancy diagnosed within 6 months; (IV) patients with poor image quality, or thrombus too small to be measured; (V) patients with syncope, hypoxemia, hemodynamic instability, or typical electrocardiogram S1Q3T3 changes. Finally, a total of 49 patients with lower-extremity DVT were included in this study. Among these patients, they were classified into two groups based on the results of CTPA: the “DVT-only group” and the “DVT-PE group”.
CT protocol
DE-CTV data were acquired using a third-generation dual-source CT scanner (SOMATOM Force; Siemens Healthcare, Germany) operating in the DE scan mode with the following scan parameters: tube A voltage 80 kV yielding 120 mA, tube B with tin (Sn) filter, tube B voltage 150 kV yielding 67 mA, gantry rotation speed 0.25 s per rotation, collimation 2 mm × 96 mm × 0.6 mm, and pitch 1.00. And the effective current was automatically adjusted to each patient’s body size. The venous spectral images were acquired with a delay of 150–180 seconds after injecting 120 mL of non-ionic iodinated contrast agent (iodixanol) and 30 mL of normal saline at a rate of 4 mL/s through the antecubital vein. Bolus tracking was employed for CTPA, where a region of interest (ROI) was placed on the main pulmonary trunk. Images were acquired 6 s after the CT attenuation surpassed 100 Hounsfield units (HU).
DE-CTV parameters acquisition
The analysis was performed using the image processing software Syngo.via (Siemens Healthcare, Germany). ROIs were delineated on the fused images with a mix factor of 0.5 (50% of the low keV and 50% of the high keV spectrum). Firstly, a circular ROI (size: 5 mm2) was placed at the proximal end of the thrombus to obtain the thrombus HU values (thrombus density). To minimize partial volume effects and confounding from thrombus heterogeneity, the images were highly magnified. The 5 mm2 ROI was strictly placed in the central core of the proximal thrombus, deliberately avoiding the vessel wall, atherosclerotic calcifications, and any peripheral flowing contrast. Sagittal and coronal reformations were cross-referenced to ensure accurate placement. ROIs were automatically duplicated in the same areas to measure DE-CTV parameters. Iodine concentration (IC) and effective atomic number (Zeff) of the ROIs were obtained from the reconstructed fused images, iodine maps, and effective atomic number maps, respectively. The DE-CTV parameters of the veins 1.0 cm above the thrombus were simultaneously measured using the same size ROI. The thrombus density ratio, normalized IC (NIC, %), and normalized effective atomic number (NZeff, %) were calculated as follows: thrombus density ratio = (thrombus density/normal vein density) × 100; (thrombus IC/normal vein IC) × 100 and (thrombus Zeff/normal vein Zeff) × 100, respectively. Both NIC and NZeff are expressed as percentages (%) representing the relative ratios of the thrombus parameters normalized to the adjacent patent vein, rather than absolute values. At the same time, CT attenuation values with the same ROI were obtained on DE spectrum images (40–190 keV at 10 keV intervals), and the slope of the spectral HU curves (λHU) was calculated: λHU = (40 keVHU − 80 keVHU)/40.
Statistical analysis
Continuous variables were expressed as mean ± standard deviation (SD) or median with interquartile range (IQR), depending on their distribution. Categorical variables were presented as numbers with percentages. Baseline characteristics of the two groups were compared using the χ2 test or Fisher’s exact test for categorical variables and the t-test or Wilcoxon test for continuous variables. Univariate and multivariate logistic regression analyses were employed to identify risk factors linked to PE. Variables that showed significant associations with PE (P<0.10) in the univariate analysis were included in the multivariate analysis. To assess for multicollinearity among these highly correlated DE-CTV parameters, collinearity diagnostics were performed by calculating the Variance Inflation Factor (VIF). The results were presented as odds ratio (OR) with corresponding 95% confidence intervals (CI). To assess the predictive ability of DE-CTV parameters for PE, receiver operating characteristic (ROC) curves were constructed, and the area under the curve (AUC) was calculated. Statistical analysis was performed using R (v4.2.3; R Foundation for Statistical Computing, Vienna, Austria).
Results
Patient characteristics
Finally, 24 patients with isolated lower extremity DVT and 25 patients with concomitant lower extremity DVT and PE were included for further analysis. The flowchart of patient enrollment is shown in Figure 1. The clinical characteristics of these patients are presented in Table 1. Their mean age was 59.00±14.86 years, and 27 of 49 patients (55.1%) were female. Among these, 46 patients (93.9%) underwent endovascular treatment, while 3 patients (6.1%) received anticoagulant therapy alone: 2 patients (4.1%) due to recent major surgery, and 1 patient (2.0%) due to active gastrointestinal bleeding. There were no differences between the two groups in terms of demographics, medical history, and laboratory tests. However, compared to the DVT-only group, the DVT-PE group showed a trend toward lower mean platelet count (193.76×109/L±52.38×109/L vs. 227.46×109/L±81.28×109/L, P=0.09).
Figure 1.
Study flow chart of the patients. DVT, deep vein thrombosis; PE, pulmonary embolism; VTE, venous thromboembolism.
Table 1. Baseline characteristics of the study population.
| Characteristic | Total (n=49) | DVT-only group (n=24) | DVT-PE group (n=25) | P value |
|---|---|---|---|---|
| Male | 22 (44.9) | 8 (33.3) | 14 (56.0) | 0.19 |
| Age (years) | 59.00±14.86 | 59.25±13.96 | 58.76±15.96 | 0.91 |
| Medical history | ||||
| Hypertension | 21 (42.9) | 11 (45.8) | 10 (40.0) | 0.90 |
| Diabetes | 8 (16.3) | 5 (20.8) | 3 (12.0) | 0.46 |
| Coronary artery disease | 3 (6.1) | 3 (12.5) | 0 (0.0) | 0.11 |
| Stroke | 5 (10.2) | 4 (16.7) | 1 (4.0) | 0.19 |
| Smoke | 7 (14.3) | 2 (8.3) | 5 (20.0) | 0.42 |
| Drink | 8 (16.3) | 2 (8.3) | 6 (24.0) | 0.25 |
| Fracture | 1 (2.0) | 1 (4.2) | 0 (0.0) | 0.49 |
| Immobilization | 11 (22.4) | 5 (20.8) | 6 (24.0) | >0.99 |
| Malignancy (>6 months) | 7 (14.3) | 4 (16.7) | 3 (12.0) | 0.70 |
| Laboratory test | ||||
| WBC (×103/μL) | 7.24±1.94 | 6.82±2.08 | 7.63±1.74 | 0.15 |
| NE (%) | 68.61±9.89 | 66.32±10.26 | 70.80±9.20 | 0.11 |
| Hemoglobin (g/dL) | 124.47±19.29 | 123.25±22.62 | 125.64±15.86 | 0.67 |
| Hematocrit (%) | 36.82±5.38 | 35.86±6.06 | 37.73±4.58 | 0.23 |
| Platelet (×103/μL) | 210.27±69.47 | 227.46±81.28 | 193.76±52.38 | 0.09 |
| D-dimer (μg/mL) | 8.48 (5.03, 13.15) | 8.17 (6.93, 11.81) | 8.80 (4.35, 16.40) | 0.66 |
Data are presented as n (%), mean ± SD or median (interquartile range). DVT, deep vein thrombosis; NE, neutrophil ratio; PE, pulmonary embolism; SD, standard deviation; WBC, white blood cell.
Among the 25 patients with confirmed asymptomatic PE, the anatomical distribution based on the most proximal affected artery was: main pulmonary artery in 6 (24.0%), lobar in 7 (28.0%), segmental in 5 (20.0%), and subsegmental in 7 (28.0%). Bilateral PE was observed in 4 patients (16.0%), and unilateral PE in 21 (84.0%). Furthermore, 13 patients (52.0%) were classified as having central PE (main/lobar involvement), while 12 (48.0%) had peripheral PE (segmental/subsegmental).
DE-CTV findings
The DE-CTV images revealed a higher prevalence of accumulation in the left lower extremity (23 of 49, 46.9%), with the common iliac vein and inferior vena cava being the most commonly affected (both 14 of 49, 28.6%). However, there was no significant difference in the accumulation of thrombus between the limbs and veins in the two groups (P=0.91 and P=0.92, respectively). The thrombus density ratio in the DVT-PE group was significantly higher than that in the DVT-only group (52.81%±10.05% vs. 40.23%±11.63%, P<0.001), and similarly, the NIC (39.0%±13.1% vs. 15.2%±11.1%, P<0.001), NZeff (88.6%±3.9% vs. 83.1%±4.2%, P<0.001), and λHU (2.34; IQR: 1.92–2.45 vs. 1.54; IQR: 1.43–2.11, P<0.001) were also significantly higher in the DVT-PE group (Table 2).
Table 2. CTV findings of the study population.
| Characteristic | Total (n=49) | DVT-only group (n=24) | DVT-PE group (n=25) | P value |
|---|---|---|---|---|
| Involved limb | 0.90 | |||
| Left | 22 (44.9) | 10 (41.7) | 12 (48.0) | |
| Right | 10 (20.4) | 5 (20.8) | 5 (20.0) | |
| Bilateral | 17 (34.7) | 9 (37.5) | 8 (32.0) | |
| Involved vein | 0.93 | |||
| Superficial FV | 5 (10.2) | 3 (12.5) | 2 (8.0) | |
| Common FV | 8 (16.3) | 3 (12.5) | 5 (20.0) | |
| External IV | 9 (18.4) | 4 (16.7) | 5 (20.0) | |
| Common IV | 13 (26.5) | 7 (29.2) | 6 (24.0) | |
| IVC | 14 (28.6) | 7 (29.2) | 7 (28.0) | |
| Thrombus density ratio (%) | 46.65±12.47 | 40.23±11.63 | 52.81±10.05 | <0.001 |
| NIC (%) | 27.33±16.99 | 15.19±11.11 | 38.98±13.05 | <0.001 |
| NZeff (%) | 85.92±4.90 | 83.09±4.19 | 88.63±3.94 | <0.001 |
| λHU (%) | 1.99 (1.50, 2.38) | 1.54 (1.43, 2.11) | 2.34 (1.92, 2.45) | <0.001 |
Data are presented as n (%), mean ± SD or median (interquartile range). CTV, computed tomography venography; DVT, deep vein thrombosis; FV, femoral vein; IV, iliac vein; IVC, inferior vena cava; NIC, normalized iodine concentration; NZeff, normalized effective atomic number; PE, pulmonary embolism; SD, standard deviation; λHU, the slope of the spectral Hounsfield unit curves.
DE-CTV parameters and PE
The inter-observer and intra-observer agreements for the manual ROI measurements were evaluated across all 49 patients. For the NIC, the intra-observer intraclass correlation coefficient (ICC) was 0.923 (95% CI: 0.868–0.956), and the inter-observer ICC was 0.891 (95% CI: 0.814–0.937). For the thrombus density ratio, the intra-observer ICC was 0.888 (95% CI: 0.810–0.935), and the inter-observer ICC was 0.878 (95% CI: 0.792–0.929). Bland-Altman plots further demonstrated robust agreement with narrow limits of agreement across both observers (Figure S1).
Figure 2 shows the initial DE-CTV images and parameters for two patients with DVT (one with PE and one without). Univariate logistic regression analysis showed that the thrombus density ratio, the NIC, the NZeff, and the λHU may be associated factors with PE (Table 3). Then, these four factors were entered in the multivariate logistic regression analysis, and no significant multicollinearity was observed (all VIF <5). However, in multivariate logistic regression analysis, only the NIC was determined as an independent risk factor for PE (OR: 1.25, 95% CI: 1.09–1.53; P=0.009). The area under the ROC curve for the thrombus density ratio was 0.813 (95% CI: 0.689–0.938; P<0.001), and the area under the ROC curve for the NIC, NZeff and λHU were 0.920 (95% CI: 0.846–0.994; P<0.001), 0.837 (95% CI: 0.724–0.950; P<0.001), and 0.799 (95% CI: 0.672–0.926; P<0.001), suggesting that the NIC may be more discriminating than the other DE-CTV parameters for the manifestation of PE (Figure 3). In this exploratory analysis, a NIC cutoff value of 24.4% was identified as an optimal threshold to indicate PE, with a sensitivity of 79.2% (95% CI: 59.3–93.2%) and a specificity of 96.0% (95% CI: 78.9–99.9%).
Figure 2.
Initial DE-CTV in two patients with DVT. CT pulmonary arteriography divided the two patients into the DVT-only group (A) and the DVT-PE group (B). From left to right per row: the reconstructed fused images, iodine maps, effective atomic number maps, and the spectral HU curves derived from DE-CTV. Patient B showed a significantly higher thrombus density ratio, NIC (%), NZeff (%), and λHU than that of patient A. CT, computed tomography; DE-CTV, dual-energy computed tomography venography; DVT, deep vein thrombosis; HU, Hounsfield unit; IC, iodine concentration; NIC, normalized iodine concentration; NZeff, normalized effective atomic number; PE, pulmonary embolism; λHU, the slope of the spectral Hounsfield unit curves.
Table 3. Univariate and multivariate analysis for predictors of pulmonary embolism.
| Variables | Univariate | Multivariate | |||
|---|---|---|---|---|---|
| OR (95% CI) | P value | OR (95% CI) | P value | ||
| Male | 2.55 (0.81–8.41) | 0.11 | |||
| Age | 1.00 (0.96–1.04) | 0.91 | |||
| WBC | 1.26 (0.93–1.74) | 0.15 | |||
| NE | 1.05 (0.99–1.12) | 0.12 | |||
| Hemoglobin | 1.01 (0.98–1.04) | 0.66 | |||
| Hematocrit | 1.07 (0.96–1.20) | 0.23 | |||
| Platelet | 0.99 (0.98–1.00) | 0.09 | |||
| D-dimer | 1.01 (0.95–1.08) | 0.72 | |||
| Thrombus density ratio | 1.11 (1.05–1.20) | 0.002 | 0.92 (0.81–1.04) | 0.20 | |
| NIC | 1.20 (1.10–1.35) | <0.001 | 1.25 (1.09–1.53) | 0.009 | |
| NZeff | 1.39 (1.17–1.72) | 0.001 | 0.98 (0.74–1.30) | 0.87 | |
| λHU | 11.90 (2.93–66.81) | 0.002 | 3.60 (0.44–55.72) | 0.29 | |
CI, confidence interval; NE, neutrophil ratio; NIC, normalized iodine concentration; NZeff, normalized effective atomic number; OR, odds ratio; WBC, white blood cell; λHU, the slope of the spectral Hounsfield unit curves.
Figure 3.
ROC curve analysis to compare the predictive ability of the thrombus density ratio with that of the DE-CTV parameters. ROC curve analysis generated the best area under the ROC curve of 0.920 for the NIC (95% CI: 0.846–0.994; P<0.001), which suggested that the NIC may be the most discriminative imaging biomarker for predicting a synchronous PE. AUC, area under the curve; CI, confidence interval; DE-CTV, dual-energy computed tomography venography; NIC, normalized iodine concentration; NZeff, normalized effective atomic number; PE, pulmonary embolism; ROC, receiver operating characteristic.
Discussion
This study included DVT patients without overt PE features undergoing concurrent CTPA, evaluating the differences in clinical information and DE-CTV findings between the DVT-only and DVT-PE groups. The results indicate that the thrombus density ratio, the NIC, the NZeff, and the λHU were significantly higher in the DVT-PE group than in the DVT-only group. Univariate and multivariate logistic regression analyses showed that the above DE CTV parameters might be useful as imaging predictors of PE. In our exploratory ROC analysis, a NIC value greater than 24.4% was found to be the optimal cut-off point, with a sensitivity of 79.2% (95% CI: 59.3–93.2%) and a specificity of 96.0% (95% CI: 78.9–99.9%). Additionally, the AUC of the NIC was 0.920 (95% CI: 0.846–0.994), which was higher than the AUCs of the thrombus density ratio and the NZeff, which were 0.813 and 0.837, respectively. These results suggest that the NIC may be a more accurate predictor of whether DVT is complicated by a PE.
Thrombus composition changes as the thrombus ages, and different compositions vary in CT HU value (10). Several studies have demonstrated that thrombi rich in erythrocytes exhibit higher HU values, while those rich in platelets exhibit lower HU values (18,19). Jang et al. found that erythrocyte-rich arterial thrombus showed superior sensitivity to thrombolytic agents compared to platelet-rich thrombus (20). Additionally, studies have demonstrated that erythrocyte-rich thrombus is significantly associated with successful reperfusion of occluded vessels in stroke patients, whereas platelet-rich thrombus is associated with a lower rate of successful reperfusion (21,22). Similar to but different from arterial thrombus, venous thrombus has also been found to have a dynamic structure consisting mainly of erythrocytes and fibrin-platelet layers, with activated platelets and white blood cells randomly distributed throughout the thrombus structure (23). Interestingly, a study by Aludin et al. found that spectral CT can distinguish between blood clots with different levels of red blood cells and fibrin (24). Subsequently, a study by Jeong et al. suggested that the thrombus density ratio may be a more reliable factor than the Wells score for the manifestation of PE (12). Recently, the study by Jung et al. showed that both DVT density and thrombus density ratio have good predictive value for PE (AUC =0.737, 95% CI: 0.665–0.809, P<0.001, and AUC =0.704, 95% CI: 0.629–0.779, P<0.001, respectively) (13). In our study, ROC analysis similarly showed a good AUC value for the thrombus density ratio (AUC =0.813, 95% CI: 0.689–0.938, P<0.001), but it was not identified as a significant factor in multivariate logistic regression analysis (OR =0.92, 95% CI: 0.81–1.04, P=0.20).
DE CT is now widely used in vascular imaging, where it can improve image quality, reduce contrast and radiation exposure, and increase diagnostic accuracy (16,25). DE CT acquires attenuation data at two different energy levels and reconstructs the data using a variety of energy-selective and material-selective algorithms to achieve material decomposition (15,26). Recently, a study by Gertz et al. found that DE CT parameters can provide additional diagnostic information to help distinguish between acute pulmonary embolism and CTEPH (17). Their results revealed that DE parameters, especially the CT density measured in IC images, may serve as an imaging biomarker for identifying erythrocyte-rich thrombus and fibrin/platelet-rich thrombus (AUC =0.94, 95% CI: 0.87–0.98). Therefore, we hypothesized that DE CTV parameters might be useful for identifying DVT that is prone to dislodging the pulmonary artery. To maintain the homogeneity of our cohort and minimize confounding factors on DE measurements, we strictly applied clinical exclusion criteria. Patients with iliac vein compression syndrome were excluded because the chronic mechanical obstruction and altered local hemodynamics can severely confound contrast permeability and NIC calculations. Similarly, we excluded patients with recent (<6 months) active malignancies because acute cancer-associated thrombosis is characterized by extreme systemic hypercoagulability, which distinctively alters thrombus composition. Patients with remote (>6 months) and stable malignancies were included as their baseline coagulation profile is more akin to the general population. In this study, we found that the NIC was the most effective parameter for identifying patients with PE (AUC =0.920, 95% CI: 0.846–0.994; P<0.001). The study results suggest that DE-CTV parameters, particularly the NIC, could serve as a predictor for PE in the lower extremity DVT patients.
In this study, we compared several DE CT-derived parameters for the characterization of PE, including thrombus density ratio, NZeff, and NIC. It is important to note that NIC represents the IC within the thrombus itself, thus, NIC values are influenced by the thrombus structure, composition, and permeability to the contrast agent (15,24). By comparison, the thrombus density ratio reflects the attenuation of the thrombus relative to adjacent contrast-enhanced vessels on conventional CT images, predominantly determined by the red blood cell content of the thrombus (10). NZeff reflects the effective atomic number within an ROI, influenced by both IC and the material composition of the thrombus (27). Among these image parameters, NIC may provide more sensitive quantitative information on thrombus composition and permeability characteristics, potentially indicating thrombus age or structural organization. Clinical Scores, such as Wells or Geneva, which estimate pre-test probability based on patient history, symptoms, and risk factors (3). Thus, the established fact is that in the circumstance of NIC predicted performance evaluation, the established fact is that DVT exists, and then the occurrence of PE is predicted based on the parameters of the thrombus. Clinical scoring systems are mostly based on clinical variables, divided into different probabilities of PE occurrence. Therefore, comparing the two, it seems that each index has its own unique advantages and disadvantages. In addition, this study is a preliminary exploration of the predictive function of NIC, and further large-scale data is needed to confirm our results. While the current exploratory study focuses primarily on establishing the independent radiological value of DE-CTV parameters, we acknowledge that combining these imaging metrics with a simple clinical baseline model (incorporating accessible variables such as age, D-dimer levels, and platelet counts) is essential for true clinical translation. In future large-scale, prospective studies, it will be critical to rigorously evaluate the incremental utility of adding NIC to established clinical risk scores by calculating the AUC, Net Reclassification Improvement, and Integrated Discrimination Improvement. We believe such combined clinical-radiological models could yield more personalized and highly accurate risk stratification algorithms for managing DVT patients.
There were some limitations in our study. First, this study was explicitly designed as an exploratory biomarker analysis rather than a study to develop a definitive multivariable clinical prediction model. Because of the limited number of PE events (n=25), the multivariable regression risks instability. Furthermore, the proposed NIC cutoff of 24.4% and its performance metrics are strictly exploratory; they risk being overly optimistic because internal validation (e.g., bootstrapping) and calibration plots were not performed. Future large-scale, prospective studies incorporating independent validation cohorts are required to construct and validate a formal predictive model. Second, our single-center retrospective study had a small sample size, which may limit the validity of the estimates and make the results unreliable. At the same time, it’s possible to underestimate many other factors, such as initial platelet counts. Therefore, future studies with larger sample sizes are necessary to validate the findings. Third, the manual ROI placement, despite our strict protocol, is subject to selection bias as it relies on the observer’s visual judgment of the most representative thrombus area. The 5 mm2 ROI may not fully capture the heterogeneity of the entire thrombus burden. The post-processing software must be used to access the DE CTV parameters, and the ROI should be manually located and outlined. This process typically takes between 5 and 10 minutes. Thus, there is a need for machine learning methods that can automatically outline the ROI and obtain the corresponding parameters. Fourth, we have only investigated common DE parameters. These parameters may have better diagnostic efficacy if combined with radiomics that can extract more imaging features. Fifth, we did not evaluate patients with DVT in the popliteal and calf veins, which would have led to selection bias and affected the validity of our results. Finally, a non-contrast-enhanced CT was not performed on the patient due to concerns about excessive radiation exposure, which resulted in the loss of important information.
Conclusions
In conclusion, our exploratory study suggests that DE-CTV parameters, particularly the NIC, may serve as a potential imaging indicator for asymptomatic PE in patients with lower extremity DVT. Furthermore, an exploratory NIC cutoff greater than 24.4% might help identify patients at higher risk of having a concurrent PE; however, this threshold requires future external validation.
Supplementary
The article’s supplementary files as
Acknowledgments
None.
Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments, and approved by the Ethics Committee of The First Affiliated Hospital with Nanjing Medical University (protocol code: 2024-SR-517). As this was a retrospective study, informed consent was waived.
Footnotes
Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1-0250/rc
Funding: This work was supported by Medical Research Project of Jiangsu Provincial Health Commission (No. K2023013).
Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1-0250/coif). The authors have no conflicts of interest to declare.
Data Sharing Statement
Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1-0250/dss
References
- 1.ISTH Steering Committee for World Thrombosis Day . Thrombosis: a major contributor to the global disease burden. J Thromb Haemost 2014;12:1580-90. 10.1111/jth.12698 [DOI] [PubMed] [Google Scholar]
- 2.Wendelboe AM, Raskob GE. Global Burden of Thrombosis: Epidemiologic Aspects. Circ Res 2016;118:1340-7. 10.1161/CIRCRESAHA.115.306841 [DOI] [PubMed] [Google Scholar]
- 3.Konstantinides SV, Meyer G, Becattini C, et al. 2019 ESC Guidelines for the diagnosis and management of acute pulmonary embolism developed in collaboration with the European Respiratory Society (ERS). Eur Heart J 2020;41:543-603. 10.1093/eurheartj/ehz405 [DOI] [PubMed] [Google Scholar]
- 4.Ceriani E, Combescure C, Le Gal G, et al. Clinical prediction rules for pulmonary embolism: a systematic review and meta-analysis. J Thromb Haemost 2010;8:957-70. 10.1111/j.1538-7836.2010.03801.x [DOI] [PubMed] [Google Scholar]
- 5.Anderson DR, Kahn SR, Rodger MA, et al. Computed tomographic pulmonary angiography vs ventilation-perfusion lung scanning in patients with suspected pulmonary embolism: a randomized controlled trial. JAMA 2007;298:2743-53. 10.1001/jama.298.23.2743 [DOI] [PubMed] [Google Scholar]
- 6.van der Hulle T, Dronkers CE, Klok FA, et al. Recent developments in the diagnosis and treatment of pulmonary embolism. J Intern Med 2016;279:16-29. 10.1111/joim.12404 [DOI] [PubMed] [Google Scholar]
- 7.Mazzolai L, Ageno W, Alatri A, et al. Second consensus document on diagnosis and management of acute deep vein thrombosis: updated document elaborated by the ESC Working Group on aorta and peripheral vascular diseases and the ESC Working Group on pulmonary circulation and right ventricular function. Eur J Prev Cardiol 2022;29:1248-63. 10.1093/eurjpc/zwab088 [DOI] [PubMed] [Google Scholar]
- 8.Guman NAM, van Dooremaal BM, Beenen LF, et al. Diagnostic accuracy of computed tomography venography compared with ultrasonography for detection of peripherally inserted central venous catheter-related thrombosis in patients with cancer: the DETECT study. J Thromb Haemost 2026;24:180-9. 10.1016/j.jtha.2025.08.008 [DOI] [PubMed] [Google Scholar]
- 9.Vedantham S, Grassi CJ, Ferral H, et al. Reporting standards for endovascular treatment of lower extremity deep vein thrombosis. J Vasc Interv Radiol 2009;20:S391-408. 10.1016/j.jvir.2009.04.034 [DOI] [PubMed] [Google Scholar]
- 10.Kirchhof K, Welzel T, Mecke C, et al. Differentiation of white, mixed, and red thrombi: value of CT in estimation of the prognosis of thrombolysis phantom study. Radiology 2003;228:126-30. 10.1148/radiol.2273020530 [DOI] [PubMed] [Google Scholar]
- 11.Ding YH, Abbasi M, Michalak G, et al. Characterization of thrombus composition with multimodality CT-based imaging: an in-vitro study. J Neurointerv Surg 2021;13:738-40. 10.1136/neurintsurg-2020-016799 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Jeong MJ, Kwon H, Noh M, et al. Relationship of Lower-extremity Deep Venous Thrombosis Density at CT Venography to Acute Pulmonary Embolism and the Risk of Postthrombotic Syndrome. Radiology 2019;293:687-94. 10.1148/radiol.2019190358 [DOI] [PubMed] [Google Scholar]
- 13.Jung JH, Kim JK, Kim T, et al. Clinical value of deep vein thrombosis density on lower-extremity CT venography: prediction of pulmonary thromboembolism. Curr Med Imaging 2023. [Epub ahead of print]. doi: . 10.2174/1573405620666230405104312 [DOI] [PubMed] [Google Scholar]
- 14.Yu MH, Jung JH, Kim T, et al. Thrombus volume and Hounsfield unit density as a predictor of pulmonary embolism in patients with deep vein thrombosis. Acta Radiol 2023;64:2198-204. 10.1177/02841851231165921 [DOI] [PubMed] [Google Scholar]
- 15.McCollough CH, Leng S, Yu L, et al. Dual- and Multi-Energy CT: Principles, Technical Approaches, and Clinical Applications. Radiology 2015;276:637-53. 10.1148/radiol.2015142631 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Machida H, Tanaka I, Fukui R, et al. Dual-Energy Spectral CT: Various Clinical Vascular Applications. Radiographics 2016;36:1215-32. 10.1148/rg.2016150185 [DOI] [PubMed] [Google Scholar]
- 17.Gertz RJ, Lennartz S, Kaya K, et al. Dual-layer dual-energy CT characterization of thrombus composition in acute pulmonary embolism and chronic thromboembolic pulmonary hypertension. Int J Cardiovasc Imaging 2025;41:303-14. 10.1007/s10554-024-03309-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Bacigaluppi M, Semerano A, Gullotta GS, et al. Insights from thrombi retrieved in stroke due to large vessel occlusion. J Cereb Blood Flow Metab 2019;39:1433-51. 10.1177/0271678X19856131 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Fitzgerald S, Mereuta OM, Doyle KM, et al. Correlation of imaging and histopathology of thrombi in acute ischemic stroke with etiology and outcome. J Neurosurg Sci 2019;63:292-300. 10.23736/S0390-5616.18.04629-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Jang IK, Gold HK, Ziskind AA, et al. Differential sensitivity of erythrocyte-rich and platelet-rich arterial thrombi to lysis with recombinant tissue-type plasminogen activator. A possible explanation for resistance to coronary thrombolysis. Circulation 1989;79:920-8. [DOI] [PubMed] [Google Scholar]
- 21.Hashimoto T, Hayakawa M, Funatsu N, et al. Histopathologic Analysis of Retrieved Thrombi Associated With Successful Reperfusion After Acute Stroke Thrombectomy. Stroke 2016;47:3035-7. 10.1161/STROKEAHA.116.015228 [DOI] [PubMed] [Google Scholar]
- 22.Douglas A, Fitzgerald S, Mereuta OM, et al. Platelet-rich emboli are associated with von Willebrand factor levels and have poorer revascularization outcomes. J Neurointerv Surg 2020;12:557-62. 10.1136/neurintsurg-2019-015410 [DOI] [PubMed] [Google Scholar]
- 23.Lippi G, Favaloro EJ. Venous and Arterial Thromboses: Two Sides of the Same Coin? Semin Thromb Hemost 2018;44:239-48. 10.1055/s-0037-1607202 [DOI] [PubMed] [Google Scholar]
- 24.Aludin S, Horr A, Schmill LP, et al. Experimental Characterisation of Differently Composed Thrombus Entities with Spectral-Detector-CT. Neurol Int 2026;18:38. 10.3390/neurolint18020038 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Hamid S, Nasir MU, So A, et al. Clinical Applications of Dual-Energy CT. Korean J Radiol 2021;22:970-82. 10.3348/kjr.2020.0996 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Agostini A, Borgheresi A, Mari A, et al. Dual-energy CT: theoretical principles and clinical applications. Radiol Med 2019;124:1281-95. 10.1007/s11547-019-01107-8 [DOI] [PubMed] [Google Scholar]
- 27.Goo HW, Goo JM, Dual-Energy CT: New Horizon in Medical Imaging. Korean J Radiol 2017;18:555-69. 10.3348/kjr.2017.18.4.555 [DOI] [PMC free article] [PubMed] [Google Scholar]