Original Contribution

Predicting Iliac Vein Compression With Computed Tomography Angiography and Venography: Correlation With Intravascular Ultrasound

Nicolas W. Shammas, MD, MS;  Gail A. Shammas, BS, RN;  Sue Jones-Miller, MS;  Qais Radaideh, MD;  Allyson R. Winter, DO;  Andrew N. Shammas, BS;  Istvan Z. Kovach, RA;  Bassel Bou Dargham, MD;  Ghassan E. Daher, MD;  Rayan Jo Rachwan, MD;  W. John Shammas, BS;  Waleed Omar, BS;  Aman Manazir, RA;  Srikanth Kasula, MD

Nicolas W. Shammas, MD, MS;  Gail A. Shammas, BS, RN;  Sue Jones-Miller, MS;  Qais Radaideh, MD;  Allyson R. Winter, DO;  Andrew N. Shammas, BS;  Istvan Z. Kovach, RA;  Bassel Bou Dargham, MD;  Ghassan E. Daher, MD;  Rayan Jo Rachwan, MD;  W. John Shammas, BS;  Waleed Omar, BS;  Aman Manazir, RA;  Srikanth Kasula, MD

Abstract: Background. Intravascular ultrasound (IVUS) is considered the gold standard in diagnosing common iliac vein (CIV) compression. The presence of >50% surface area reduction by IVUS is considered significant compression by most operators. Thus, we evaluated the role of computed tomography angiography (CTA) and venography in diagnosing CIV compression when compared to IVUS. Methods. All patients who underwent CTA of the pelvis with venous filling phase, IVUS, and venography within a few weeks apart to evaluate for symptomatic CIV compression from one cardiovascular practice were retrospectively reviewed. Quantitative vascular analysis was performed on all images obtained to determine (1) percent stenosis (PS) by venogram; and (2) minimal lumen area (MLA) and PS by CTA and IVUS at the compression site (using ipsilateral distal CIV as reference area). Spearman’s rank correlation, paired t-tests, or signed rank tests were performed as appropriate to compare between values of MLA and PS among the three different imaging modalities. Results. A total of 96 patients were included (62.5% females; mean age, 62.3 ± 14.8 years). A significant correlation was found between MLA-CTA and MLA-IVUS (Spearman’s rho, 0.27; P=.01) and PS-CTA and PS-IVUS (Spearman’s rho, 0.327; P<.01). A significant correlation was also found between PS-venogram and PS-IVUS (Spearman’s rho, 0.471; P<.001). MLA-CTA and MLA-IVUS had a median difference of +41 mm2 (95% CI, 25.0-57.5; P<.001) whereas PS-CTA and PS-IVUS were not statistically different (median difference, -5.6 mm2; 95% CI, -12.2 to 0.7). Furthermore, PS-IVUS and PS-venogram had a median difference of +15.2% (95% CI, 10.4-20.1; P<.001). Conclusion. PS-venogram correlates with PS-IVUS, but venogram underestimates the PS by an average of 15.2%. In contrast, PS-CTA and PS-IVUS are not statistically different despite an over-estimation of MLA by CTA when compared to IVUS. Therefore, we conclude that PS-CTA and not PS-venogram can be used to predict PS on IVUS.  

J INVASIVE CARDIOL 2018;30(12):452-455.

Key words: balloon angioplasty, computed tomography angiography, deep vein, iliac vein compression, intravascular ultrasound, venogram, venous stenting

Endovascular stenting of iliofemoral veins has shown to be effective in reducing the symptoms of swelling and pain associated with iliac vein compression (ILVC).1-6 Intravascular ultrasound (IVUS) is currently considered the gold standard in diagnosing ILVC and venography has consistently under-appreciated the severity of the compression. Computed tomography angiography (CTA) emerged as an important non-invasive imaging modality to evaluate the degree of ILVC prior to committing the patient to an invasive procedure. The correlation between CTA and IVUS needs to be further defined. We therefore report data from a single cardiovascular group on the role of CTA and venography in diagnosing ILVC compression when compared to IVUS. A scoring system was also derived using CTA findings and clinical variables to predict more accurately ILVC on IVUS. 


All patients from a single practice who underwent iliofemoral vein compression treatment at our endovascular lab from September 1, 2013 to May 30, 2017 were identified. Medical records were all retrospectively reviewed for demographic, clinical, and procedural variables. Quantitative analyses were performed on available CTA, IVUS, and venographic images; the minimal luminal area (MLA) – or diameter for venograms – at the compression site and ipsilateral reference area were measured,7-9 and the percent stenosis (PS) was calculated. The study was approved by the institutional review board of the Genesis Health System. A waiver of consent was obtained from each patient, and the review adhered to all HIPAA regulations. The study was internally audited by a clinical research associate who reviewed 100% of all records and data entry. 

Clinical data collected included age, sex, race, presence of diabetes mellitus, and history of prior ablation or vein stripping to the superficial veins, patient symptoms of lower-extremity discomfort, venous claudication, swelling or hyperpigmentation in the lower extremities (unilateral vs bilateral), claudication, ankle brachial index of the affected limb, chronic renal insufficiency (defined as creatinine >1.5 mg/dL or creatinine clearance <50 mL/min), dialysis, venous ulcerations, venous reflux (superficial and deep) as evaluated by duplex venous ultrasound, deep vein thrombosis, and lower-extremity cellulitis. IVUS data included severity of the compression (MLA and PS) and MLA following postdilation of the stent. Venography data included percent diameter stenosis of the compressed venous segment, and the presence of collaterals. Patients were only excluded if imaging studies were incomplete or uninterpretable. A significant compression on IVUS (Figure 1) was considered to be present if the PS using MLA (at compression site compared to ipsilateral distal common iliac vein) was ≥50%. 

Imaging acquisition. A 64-multidetector CT (Toshiba) with CT venography was used to image the iliac veins and perform measurements. Images were processed on a Vital workstation, and three-dimensional volume rendering, multiplanar reformation (MPR), and maximum intensity projection (MIP) images were obtained. The MLA at the compression site was measured on transverse view. The ipsilateral non-compressed iliac vein was used as the reference and MLA at this level was also measured. PS was then calculated as (MLA reference – MLA compression/MLA reference) x 100. 

The venogram was performed using an 8 Fr sheath in the left common femoral vein. Anteroposterior projection and ipsilateral lateral projection were obtained. The severity of the lesion as seen by venography was determined quantitatively from the lateral projection when this was available. The presence of collaterals was noted. 

IVUS was performed with peripheral IVUS catheters (Volcano). This was done via the 8 Fr ipsilateral sheath, and imaging started from the distal inferior vena cava (IVC) to the ipsilateral common femoral vein. MLA at compression site was measured. A reference MLA of the ipsilateral non-compressed iliac vein was also measured to determine PS at the compression site. The presence of spurs or fibrosis was noted. 

Endpoints. The primary endpoint of the study is the correlation between CTA-MLA, CTA-PS, or venography-PS with IVUS-MLA or IVUS-PS. A secondary analysis was to calculate a score based on a logistic regression model using CTA and clinical variables that would accurately predict significant ILVC on IVUS. 

Statistical analysis. Analysis was performed per patient. Descriptive analysis on all variables was done. Wilcoxon signed rank test was used on median differences and confidence intervals (CIs). The “gold standard” was defined as the percent of stenosis by IVUS ≥50%. Spearman’s rank correlation was used to describe the association between CTA and IVUS for the PS and MLA variables and the association between the gold standard and the variables collected (demographics, clinical history, and vein history). The significant correlations were then used in binary logistic regression modeling. Several models were considered utilizing the variables with significant correlations and exploring potential interactions between the variables by using backward elimination method. The best model was selected based on individual significance of variables, overall model significance (P<.05), and using the Hosmer & Lemshow test to evaluate goodness of fit (P>.05). Receiver operating characteristic (ROC) curve was initially performed (Figure 2) with the model to analyze sensitivity and 100-specificity. The model was significant, with an area under the ROC curve of 0.739 (95% CI, 0.628-0.849) with a P<.01. A single cut-off score ≥0.60 would result in a sensitivity of 75.4% and 100-specificity of 40.0%. Further analysis was then done simulating values for CTA-MLA from 20 mm2 to 400 mm2 and ages from 15 years to 90 years. The suggested new cut-off scores are a result of this simulation (Table 3). Minitab 17 and SPSS software were used for this analysis. 


A total of 96 patients were included. Table 1 describes the demographics and clinical variables. The mean age was 62.3 ± 14.8 years and 62.5% were females. Table 2 describes CTA, venogram, and IVUS measurements. A significant correlation was found between MLA-CTA and MLA-IVUS (Spearman’s rho, 0.27; P=.01) and PS-CTA and PS-IVUS (Spearman’s rho, 0.327; P<.01). A significant correlation was also found between PS-venogram and PS-IVUS (Spearman’s rho, 0.471; P<.001). MLA-CTA vs MLA-IVUS had a median difference of +41 mm2 (95% CI, 25.0-57.5; P<.001) whereas PS-CTA vs PS-IVUS was not statistically different (median difference, -5.6 mm2; 95% CI, -12.2 to 0.7). Furthermore, PS-IVUS vs PS-venogram had a median difference of +15.2% (95% CI, 10.4-20.1; P<.001).

Logistic modeling indicated that age and MLA-CTA were the strongest predictors of ILVC. Various cut-off limits for significant MLA-CTA were tested in the model, and an MLA of ≤100 mm2 appeared to be the strongest predictor of ILVC. The score below was generated from the best probability fit derived from the logistic regression analysis. 

Table 3 shows the score and age ranges obtained from the best model. The positive predictive value and negative predictive value using these scores and age ranges were 99.5% and 90%, respectively, with an accuracy of 91.1%. An example of how to interpret the data in Table 3 is as follows: in row 1, for a symptomatic patient between the age of 15 and 19 years, a score of ≥0.97 correctly predicts 91.9% of the time that a compression of ≥50% will be present on IVUS. If the score is <0.97, then 99.4% of the time there is no compression present on IVUS. 


ILVC is a prevalent condition occurring in 20%-25% of people, and varies in severity from mild to severe, with or without associated symptoms.4 The gold standard diagnosis has been IVUS, which requires an invasive evaluation. Venography under-estimates the severity of compression by about 15%, confirming prior published data.10 PS on a CTA appeared to be comparable to what we have seen on IVUS. However, several limitations exist in determining PS on a CTA. First, the ipsilateral reference CIV may have significant prestenotic dilation, and this may exaggerate the compression severity. On the other hand, the compression may well involve the entire CIV in certain variants of May-Thurner, and therefore compression severity could be under-estimated. Also, it is not infrequent to have difficulty measuring the distal left CIV reference because of angulation, leading to inaccurate measurements. Although the contralateral CIV has been used as a reference vessel in situations where an ipsilateral CIV reference could not be obtained, these measurements assume that the CIVs bilaterally are equal in size, which may be erroneous. 

Using more predictable measurements on CTA, coupled with clinical criteria to improve the positive predictive value and negative predictive value of the test, will likely improve the accuracy of CTA in identifying patients with significant ILVC. MLA is a relatively easy measurement that can be obtained from the CTA during the venous filling phase. In this study, an MLA of ≤100 mm2 seems to predict the presence of a significant compression in a symptomatic population as seen on IVUS (>50%). This cut-off limit seems also to be influenced by the patient’s age on the regression model. By combining these two variables, a range of scores was obtained that can accurately predict ILVC.  A higher score than the cut-off score derived from the model would indicate that ILVC is likely to be present on IVUS. This scoring system is currently being tested prospectively. 

ILVC is commonly found on CTA, particularly in patients with deep vein thrombosis.11,12 ILVC decreases with age.12 Similarly, in our study, age was an independent predictor of symptomatic ILVC and was incorporated as part of the scoring system. The presence of an MLA <100 mm2 and age were the only predictors of significant ILVC as seen on IVUS. 

Study limitations. This study is retrospective and selection bias is a possibility. However, we included all symptomatic patients, and images were quantitatively interpreted. Therefore, these data should not be extrapolated to asymptomatic ILVC, which is generally treated conservatively and thus a CTA is unlikely to be performed for the purpose of diagnosis. 


CTA can be a reliable tool to determine PS as seen on IVUS, and the combination of MLA on CTA and age can be used to predict accurately the presence of ILVC on IVUS. 


1.    McMurrich JP. The occurrence of congenital adhesions in the common iliac veins and their relation to thrombosis of the femoral and iliac veins. Am J Med Sci. 1908;135:342-346. 

2.    May R, Thurner J. The cause of the predominantly sinistral occurrence of thrombosis of the pelvic veins. Angiology. 1957;8:419-427.

3.    Cockett FB, Thomas ML. The iliac compression syndrome. Br J Surg. 1965;52:816-821. 

4.    Kibbe MR, Ujiki M, Goodwin AL, Eskandari M, Yao J, Matsumura J. Iliac vein compression in an asymptomatic patient population. J Vasc Surg. 2004;39:937-943.

5.    Abboud G, Midulla M, Lions C, et al. “Right-sided” May-Thurner syndrome. Cardiovasc Intervent Radiol. 2010;33:1056-1059.

6.    Kölbel T, Lindh M, Akesson M, Wassèlius J, Gottsäter A, Ivancev K. Chronic iliac vein occlusion: midterm results of endovascular recanalization. J Endovasc Ther. 2009;16:483-491.

7.    Oguzkurt L, Ozkan U, Tercan F, Koç Z. Ultrasonographic diagnosis of iliac vein compression (May-Thurner) syndrome. Diagn Interv Radiol. 2007;13:152-155. 

8.    Oguzkurt L, Tercan F, Pourbagher MA, et al. Computed tomography findings in 10 cases of iliac vein compression (May-Thurner) syndrome. Eur J Radiol. 2005;55:421-425. 

9.    Neglen P, Raju S. Intravascular ultrasound scan evaluation of the obstructed vein. J Vasc Surg. 2002;35:694-700.

10.    Gagne PJ, Tahara RW, Fastabend CP. Venography versus intravascular ultrasound for diagnosing and treating iliofemoral vein obstruction. J Vasc Surg Venous Lymphat Disord. 2017;5:678-668.

11.    Chung JW, Yoon CJ, Jung SI, et al. Acute iliofemoral deep vein thrombosis: evaluation of underlying anatomic abnormality by spiral CT venography. J Vasc Interv Radiol. 2004;15:249-256.

12.    Nazzal M, Elfedaly M, Qu W, Kazan V, Abbas J, Zelenock G. Left common iliac vein compression is not uncommon CT finding. J Vasc Surg. 2012;55:298.

From the Midwest Cardiovascular Research Foundation, Davenport, Iowa.

These data were submitted for presentation at Cardiovascular Innovations 2018 in Denver, Colorado and the Midwest Cardiovascular Forum 2018, Minneapolis, Minnesota.

Disclosure: The authors have completed and returned the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr N.W. Shammas reports educational and research grants from Bard, Boston Scientific and Intact Vascular. The remaining authors have no conflicts of interest regarding the content herein.  

Manuscript submitted October 14, 2018, provisional acceptance given October 23, 2018, final version accepted October 25, 2018.

Address for correspondence: Nicolas W. Shammas, MD, MS, EJD, Midwest Cardiovascular Research Foundation, 1622 E. Lombard Street, Davenport, IA 52803. Email: shammas@mchsi.com