Venous ultrasonography has emerged as a pivotal non-invasive technique for assessing systemic congestion. One notable modality within this field is the Venous Excess Ultrasound Score (VExUS) protocol, introduced in 2020 by Beaubien-Souligny et al. [1]. This protocol facilitates the estimation of venous congestion through the evaluation of the inferior vena cava (IVC) and Doppler assessment of the hepatic veins (HV), portal vein (PV), and intrarenal veins (RV).

Initially, this scoring system was validated in post-operative cardiac surgery patients, in whom VExUS has been shown to predict the risk of organ dysfunction, particularly acute kidney injury [1]. Subsequently, Longino demonstrated a positive correlation between right atrial pressure measurements obtained via catheterization and VExUS scores [2,3].

This finding has prompted consideration of the potential utility of this tool in other conditions characterized by venous system overload, such as pulmonary embolism (PE)—the third most common cardiovascular syndrome after acute myocardial infarction and stroke [4]. The available evidence regarding the utility of VExUS in patients with PE is limited to the study by Rola et al., which evaluated a single case of cardiogenic shock due to massive pulmonary embolism in a pregnant patient presenting with a pathological VExUS score [5]. To date, no further data have been identified in the literature to clarify the role this tool might play in patients with PE.

Our objective is to analyze the ultrasonographic findings of VExUS in patients with acute pulmonary embolism and to determine whether it accurately correlates with the full spectrum of disease severity—namely, low risk, intermediate-low risk, intermediate-high risk, and high risk PE.

During the study period, 90 eligible patients were identified, of whom 9 were not evaluated for various reasons (detailed in Fig. 1). Of the remaining 81 patients, the VExUS protocol was completed in 80 participants (43.8% women, mean age 66.7±15.7 years), yielding a feasibility rate of 98.8% (95% CI: 93.3–100%).

Fig. 1.

Flowchart.

The most frequent comorbidities were hypertension (51.2%), solid tumors (23.8%), and diabetes mellitus (21.3%). The most common risk factors for PE included varicose veins (26.3%), prior immobilization (23.8%), history of venous thromboembolism (18.8%), and recent surgery (16.3%). Baseline characteristics are summarized in Table 2.

PE was diagnosed primarily via pulmonary artery CT angiography (97.5%). According to the ESC risk classification, 23 patients (28.8%) had low risk, 34 (42.5%) intermediate-low risk, 17 (21.3%) intermediate-high risk, and 6 (7.5%) high risk PE (Table 2). No patients died during hospitalization.

The inter-observer variability analysis demonstrated a good level of agreement across all measurements: the weighted Kappa coefficient was 0.80 (95% CI: 0.44–1.00) for the HV pattern, 0.62 (95% CI: 0.34–1.00) for the RV pattern, and 1.00 (95% CI: 1.00–1.00) for the global VExUS score. For continuous measurements, the intraclass correlation coefficient (ICC) was 0.88 (95% CI: 0.61–0.97) for the IVC diameter and 0.85 (95% CI: 0.59–0.98) for the PV pulsatility index.

A statistically significant ascending trend was observed in VExUS scores across ESC risk categories. All low- and intermediate-low risk patients had a VExUS score of 0. Among intermediate-high-risk patients, 6 (35.3%) had a VExUS score of 0, 6 (35.3%) a score of 1, 4 (23.5%) a score of 2, and 1 (5.9%) a score of 3. No high-risk patient had a VExUS score of 0; 2 (33.3%) had a score of 1, 3 (50%) a score of 2, and 1 (16.7%) a score of 3.

The IVC diameter was greater in the higher-risk groups, with a mean of 22.2mm in the high-risk group vs. 14.1mm in the low- and intermediate-low-risk groups. Inspiratory IVC collapse>50% was observed exclusively in the lower-risk groups (87.9% in intermediate-low and 87% in low risk), with no such finding in the higher-risk groups. Regarding the HV, the normal pattern predominated in the lower-risk groups (94.1% and 100%), while abnormal patterns were mostly observed in intermediate-high-risk participants (52.9% and 11.8%, respectively) and in high-risk participants (33.3% and 50%, respectively). The PV pulsatility index also showed a progressive increase. Finally, renal vein analysis revealed continuous flow in all low- and intermediate-low-risk patients, whereas biphasic and monophasic patterns were identified in 33.4% of the high-risk group and 23.6% of the intermediate-high-risk group, suggesting progressive systemic venous congestion related to clinical severity. The details of the findings are described in Table 3.

A progressive increase in the mean response was observed as the VExUS score increased, as shown in Table 4. The Jonckheere–Terpstra test yielded a statistically significant result (J=1020, equivalent to z=6.309; p<0.001). The effect size was estimated at 0.701, corresponding to a medium effect. Five patients underwent reperfusion therapy, with their ultrasound examinations performed after the intervention (median time to ultrasound: 5h; interquartile interval, 5–16h). After excluding these patients from the analysis, similar results were observed (J=686, z=5.30; p<0.001, effect size=0.61). Similarly, the exclusion of five patients with a diagnosis of in-hospital pulmonary embolism did not result in significant changes to the estimates (J=935, z=6.23, p<0.001; effect size=0.72).

The ROC curve demonstrated good ability to discriminate high-risk patients (AUC 0.94, 95% CI: 0.89–0.99) (Fig. 2). Based on this analysis, we determined that the optimal cut-off point was>1 point. This threshold provided a sensitivity of 100% and a specificity of 85%. In populations with a similar prevalence of high-risk PE to ours (7.5%), the negative predictive value would be 100%, and the positive predictive value would be 35.3%.

Fig. 2.

ROC curve.

In the ordinal logistic regression model, the PV pulsatility index was identified as the parameter most precisely associated with severity (OR 1.16; 95% CI: 1.06–1.26; p=0.001). HV patterns also showed a significant association, though with low precision (S<D: OR 10.0; 95% CI: 1.27–78.8; p=0.029; inverted S: OR 110.4; 95% CI: 6.45–1889.9; p=0.001). The IVC diameter and RV pattern did not demonstrate a significant association with PE severity. However, the model violated the proportional odds assumption (likelihood-ratio χ2=30.63, p=0.0007).

Therefore, the categories of the categorical variables (HV and RV patterns) were regrouped into ‘no congestion’ and ‘congestion’ (i.e., S>D vs. S<D or inverted S for HV; continuous vs. non-continuous flow for RV), and the model was re-estimated. The results were similar to those obtained with the original categories (Table 5). This model satisfied the proportional odds assumption (likelihood-ratio χ2=10.75, p=0.1498). As an exploratory analysis, adjustments were made for sex, age, and renal function. The results did not change substantially (Table 5).

To explore the incremental value of VExUS, we adopted an exploratory approach. Two binary logistic regression models were fitted using the need for reperfusion therapies as the dependent variable. The first model included ESC risk categories as independent variable, while the second model added VExUS score. After estimation, the Net Reclassification Improvement (NRI) was calculated using a 0.5 probability threshold for event occurrence. The addition of VExUS to ESC risk categories for predicting the need for reperfusion therapy resulted in a NRI of 11.1%.

This study demonstrates a significant association between the severity classification of PE and ultrasonographic findings obtained using the VExUS protocol, showing a progressive increase in venous congestion as PE severity worsens. To the best of our knowledge, this is the first study to systematically evaluate the use of the VExUS protocol in patients with PE.

In acute PE, acute right ventricular overload leads to increased systemic congestion secondary to a sudden rise in afterload and impaired right ventricle filling [6,7]. Our findings, which suggest venous congestion across all examined territories, align with the pathophysiology described by Bryce et al. [8]. One possible interpretation of our findings is that multiorgan involvement in pulmonary embolism could reflect, at least in part, a direct mechanical consequence of increased right ventricular afterload, rather than primary peripheral organ injury. If this hypothesis holds, systemic venous congestion might become apparent even before a clinically significant reduction in cardiac output occurs. Such a mechanism could potentially explain why alterations in hepatic venous congestion patterns (both hepatic and portal veins) were observed earlier than changes in the inferior vena cava. However, this interpretation is limited by the design of our study, as we did not perform repeated VExUS examinations to assess temporal trends in venous congestion. Thus, this remains hypothetical, and the pathophysiological interplay between right heart dysfunction and peripheral organ congestion in acute PE—particularly in high-risk patients—warrants further investigation, given the marked differences between PE and other clinical contexts, such as cardiac surgery, in which other studies were carried out.

The findings of our study are consistent with those described by Beaubien-Souligny et al. [1]. Although both studies were conducted in different populations (cardiac surgery vs. PE), the results consistently demonstrate that the presence of venous abnormalities on ultrasound is associated with greater hemodynamic compromise. The medium effect size (0.701) observed in this study suggests that the VExUS score is not only statistically associated with PE severity but also clinically relevant. An effect size of this magnitude indicates a substantial and meaningful relationship between venous congestion—as assessed by VExUS—and the hemodynamic impact of PE. Additionally, the results did not substantially change when excluding patients who underwent reperfusion therapy and in whom the ultrasonographic study was performed afterward. Additionally, the results did not substantially change when excluding patients who underwent reperfusion therapy and in whom the ultrasonographic study was performed afterward. In any case, it is reasonable to expect that higher-risk patients—those who required reperfusion—would have had an even worse VExUS score prior to the procedure. Additionally, the findings did not differ significantly between patients with pulmonary embolism diagnosed in an outpatient setting versus those diagnosed during hospitalization. This consistency across subgroups strengthens the internal validity of the study and suggests that the VExUS protocol may offer reliable diagnostic utility regardless of the clinical setting.

In our analysis, the portal pulsatility index was the most accurate parameter for predicting severity, unlike the findings reported by Beaubien-Souligny et al., where the diastolic pattern of the hepatic veins was the individual parameter most strongly associated with acute kidney injury [1]. These results also differ from those observed by Longino et al. [2,3] in heart failure patients, where the renal component of VExUS demonstrated one of the strongest associations with central venous pressure measured via cardiac catheterization. These discrepancies may be explained by pathophysiological differences between the two populations. In acute PE, the elevation in systemic venous pressure may be more abrupt and transient, potentially precluding the development of monophasic renal patterns typical of chronic congestion. In contrast, portal and hepatic changes—more sensitive to acute variations in venous return—may better reflect the initial hemodynamic dysfunction characteristic of pulmonary embolism.

In our study, the inferior vena cava (IVC) diameter alone did not demonstrate a significant association with PE severity, consistent with the findings of the protocol's developers [1]. This suggests that ultrasound assessment of a single venous territory is insufficient to detect clinically significant congestion, and that combining multiple venous territories is necessary to improve overall diagnostic consistency [2,3]. These findings align with a recent study of 124 heart failure patients, which found that the VExUS score provided more accurate results than individual IVC characteristics [9].

In our view, the stability of these results after adjustment for age, sex, and baseline renal function further strengthens their validity, reinforcing the robustness of the VExUS approach in this clinical context.

The feasibility of the protocol in our cohort was high (98.8%), similar to that reported in other studies [1–3,9], confirming its applicability across different clinical contexts, including patients with acute respiratory compromise. It is also important to highlight that the protocol, when performed by specifically trained personnel, appears to be highly reproducible—both in its continuous measurements and in category assignment. However, these findings are based on a limited number of observations, and it would be beneficial to have a larger sample of studies conducted by multiple operators to assess intra- and inter-observer variability.

This study introduces a novel, feasible, and non-invasive method for risk assessment in PE, complementary to existing approaches. It exhibited excellent performance in identifying high-risk patients, as demonstrated by ROC curve analysis, which indicated a high discriminative ability with both high sensitivity and specificity. Interestingly, the addition of VExUS results to the traditional ESC risk scale improved the prediction of reperfusion therapy needs by 11%. While these findings are exploratory and do not account for hard outcomes such as mortality—due to limitations in our dataset—they reinforce the potential utility of VExUS as a practical, non-invasive tool for risk stratification that could complement established markers, enhancing clinical decision-making in patients with pulmonary embolism, pending further validation in larger studies.

Key strengths include the blinding of investigators to patient risk categories, which minimized potential bias in the subjective ultrasound evaluation. Additionally, protocol implementation within the first 24h of diagnosis yielded valuable insights into PE pathophysiology before the initiation of antithrombotic therapies could alter clinical presentation. However, this analysis is not without limitations. First, the absence of direct central venous pressure measurement via invasive techniques precludes precise pathophysiological correlation between ultrasonographic findings and actual intracardiac pressures. Unlike the studies by Beaubien-Souligny et al. [1] and Longino et al. [2,3], which validated VExUS against invasive hemodynamic measurements or filling pressures, our work relies on clinical variables and does not allow for absolute quantification of the protocol's accuracy. Second, the single-center design limits the generalizability of the results. Furthermore, ultrasound assessments were performed at a single time point, without longitudinal follow-up or documentation of clinical outcomes such as organ dysfunction, acute kidney injury, or mortality. This precludes definitive conclusions about the prognostic value of VExUS in the context of PE, which has been explored in heart failure [10–13]. We must also acknowledge that our small sample size limits the precision of our estimates. Therefore, these findings should be interpreted with caution and will require confirmation in future research.

Considering these points, this study provides original evidence supporting the utility of VExUS in PE risk stratification, extending its applicability beyond classic heart failure congestion scenarios. Despite the limitations, the results support the pathophysiological validity of the protocol and justify the need for multicenter, longitudinal, and prospective studies to explore clinical outcomes—such as the development of post-pulmonary embolism syndrome—to confirm its prognostic value and potential role in guiding therapeutic decisions.

In conclusion, the assessment of venous congestion using the VExUS protocol in patients with acute PE demonstrates an ordered association with risk category and may serve as a complementary measure to current risk stratification methods.

The authors acknowledge the use of Mistral AI's Le Chat (version January 2026) for grammatical corrections and phrasing refinements in the preparation of this manuscript. All content was thoroughly reviewed and approved by the investigators to ensure accuracy, scientific rigor, and adherence to ethical standards. The use of AI-assisted tools was limited to language enhancement and did not influence the study design, data analysis, or conclusions.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

The authors declare not to have any conflicts of interest that may be considered to influence directly or indirectly the content of the manuscript.