Diagnostic Accuracy of Color Doppler Imaging in the Evaluation of Peripheral Arterial Disease
As Compared to Digital Subtraction Angiography
1Chiramel George Koshy, MD, 1Binita R. Chacko, MD, 1Shyamkumar N. Keshava, DNB, FRCR, FRANZCR, 2Sunil Agarwal, MS
From the 1Departments of Radiology and 2Vascular Surgery, Christian Medical College, Vellore, Tamil Nadu, India.
Correspondence: Dr. Chiramel George Koshy, Department of Radiology, Christian Medical College, Vellore, Tamil Nadu, India 632004. E-mail: [email protected].
Manuscript submitted November 12, 2008, provisional acceptance given December 5, 2008, accepted December 11, 2008.
Disclosure: The authors report no financial relationships or conflicts of interest regarding the content herein.
Purpose. To estimate the diagnostic accuracy of color Doppler imaging compared to digital subtraction angiography (DSA) in the assessment of patients with peripheral arterial disease (PAD). Methods and materials. Patients with a clinical diagnosis of lower limb PAD who were awaiting diagnostic DSA first underwent a color Doppler scan. The subsequent DSA was done by another radiologist unaware of the Doppler findings. Results were recorded on separate proforma. Statistical analysis. The evaluated segments were graded as normal, insignificant disease (/= 50%), and occlusion. The results were analyzed using two-way tables and the kappa statistic to look for concordance between the two modalities. Results. The study involved 41 patients and analyzed 720 arterial segments. Excellent concordance (1.0) was seen in the aortoiliac segments. Good concordance was seen in the common and external iliac segments (0.96–0.77), as well as in the common and superficial femoral segments (0.77–0.88). The popliteal segments showed lower concordance (0.66). There was only fair concordance (0.54–0.67) in the infrapopliteal segments, with relatively better results in the posterior tibial artery. The overall sensitivity ranged from 69–100%, specificity 69–100%, PPV 92–100%, and NPV from 70–100%, depending on the vascular segment evaluated. Conclusions. For identifying hemodynamically significant lesions, color Doppler was found to be as good as DSA in the aortoiliac and femoropopliteal regions. However, DSA is still required to evaluate the infrapopliteal segments.
Peripheral arterial disease (PAD) is a common problem faced by clinicians. Reduction of blood supply to a lower limb initially presents as intermittent claudication, while further restriction of flow leads to ischemic pain at rest. If not treated, trophic changes like ulceration and gangrene may occur and can result in loss of the limb. Diagnostic imaging is performed when PAD limits the patient’s lifestyle. Accurate characterization of the number, level, and severity of lesions is necessary to plan treatment. The imaging modality should provide adequate information and minimize risk and inconvenience to the patient. Most centers use color Doppler as the initial imaging modality, sometimes followed up with digital subtraction angiography (DSA).
Color Doppler versus DSA. Color Doppler imaging is a well-established modality in the assessment of PAD. It is safe, painless, noninvasive, reproducible, relatively inexpensive, and widely available as an outpatient service. It uses no ionizing radiation or contrast material. However, it is time consuming and does not provide an image of the arterial tree that can be easily visualized by the clinician. Moreover, the time taken for the study as well as its accuracy depend on the expertise of the sonologist.
DSA is reproducible, provides an easily visualized image of the arterial tree, can measure pressure gradients, and can be used for interventional treatment. However, this requires hospital admission and has risks associated with vascular access and catheterization. It involves the use of ionizing radiation and iodinated contrast agents. Because of its invasive nature and high cost, it is not suitable for screening or for follow-up purposes. Although it is considered to have a high level of objectivity, there can be considerable interobserver variation in interpretation.1–4
While angiography is a morphological study and provides information only about the vessel lumen, color Doppler imaging is both a morphological and functional study, providing information not only about the vessel wall but also hemodynamic information.
Recent advances like better post-processing capability, transducer technology, image resolution, signal strength, and spectral analysis capabilities have improved the ability of color Doppler to visualize and grade abnormalities, thus extending the scope for non-invasive assessment of PAD.
Angiography has been considered by many as the definitive investigation in the evaluation of PAD. Many comparative studies have considered color Doppler to be inferior to angiography.
However, a few studies have found that color Doppler could replace up to 97% of diagnostic arteriography of the lower limb5 and could safely and accurately guide therapeutic vascular interventions,6 thus suggesting that DSA no longer be regarded as the gold standard.
Materials and Methods
This prospective study was conducted in the department of Radiology, Christian Medical College, Vellore, India, between November 2006 and September 2007. The institutional review board gave their approval before the study began. All patients with a clinical diagnosis of PAD, and planned for diagnostic angiography of the lower limbs were consecutively inducted into the study after receiving informed consent. Exclusion criteria were uncorrectable blood dyscrasia, pregnancy, history of allergic reactions to iodinated contrast media, and inability to give informed consent.
Included patients first underwent a color Doppler scan of the arteries of the lower limbs to localize and grade the lesions. In order to eliminate interobserver variation, all Doppler studies were done by the same radiologist, who attempted to study every arterial segment with color Doppler to locate lesions, analyze the spectral waveform, and measure the peak systolic velocity (PSV). Gray-scale ultrasound was also used to identify morphological features of the vessel wall. Arterial lesions were assessed for an increase in PSV, spectral broadening, and reduction in the lumen. The PSV ratio was measured by comparing it to a proximal normal segment.
A normal segment was demonstrated by color saturation seen throughout the lumen and triphasic waveform pattern. Occlusion was diagnosed when no color saturation or Doppler waveform was demonstrable within the segment. A hemodynamically significant stenosis (> 50%) was diagnosed as PSV ratio of more than 2.2 For non-occlusive lesions, grading of the arterial segment with color Doppler was based on the PSV ratio, while in angiography it was based on the relative reduction of the width of the contrast column compared to that in the segment immediately proximal to it.
Uniplanar angiography was done on a Siemens Multistar Top DSA machine (Erlangen, Germany) by an interventional radiologist who knew the clinical details but was unaware of the Doppler findings. All patients were given intravenous sedation and local anesthesia prior to the angiogram.
All patients included in the study underwent both color Doppler and DSA studies. The time interval between the modalities was less than 24 hours. No treatment was administered in between.
DSA images were recorded as angiographic runs on a compact disc and also documented on the Picture Archiving and Communication System (PACS). In some cases, it was decided to proceed to endovascular treatment in the same session.
Data entry and analysis. The findings of color Doppler and DSA studies were entered on separate proforma containing a line diagram of the lower limb arteries, with shading of the involved segments indicating disease. PSV and flow pattern details for each arterial segment were mentioned in the case of color Doppler. The length of the obstruction, presence of collaterals, and point of distal reformation were mentioned, if any. Lesions were graded as follows: normal/insignificant (/= 50%), or occlusion (100%).
The following vascular segments were analyzed:
• Infrarenal aorta
• Common iliac artery
• External iliac artery
• Common femoral artery
• Superficial femoral artery (proximal one-third)
• Superficial femoral artery (mid one-third)
• Superficial femoral artery (distal one-third)
• Popliteal artery
• Anterior tibial artery
• Posterior tibial artery
• Common peroneal artery
The results of color Doppler and DSA studies were analyzed using two-way contingency tables, and the kappa statistic, with DSA as the gold standard. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were calculated segment-wise for the ability of color Doppler to predict hemodynamically significant involvement (>/= 50% stenosis and occlusion).
The study involved 41 patients (38 men, 3 women). Though 72 limbs with 837 individual arterial segments were evaluated using each modality, only 720 segments (86%) were available for comparison, being visualized on both color Doppler and DSA. The number of diseased segments evaluated was 142. Omnipaque 300 mg/ml (GE Healthcare/Amersham Health, Princeton, New Jersey) was used as the angiographic contrast agent in all patients except for two, who had reduced renal function requiring the use of CO2 as the contrast agent.
Of the patients, 60% were between 40 and 60 years of age. Most of our patients had atherosclerosis, while two had acute thrombosis and one had distal embolic disease due to a large iliac artery aneurysm. The superficial femoral arteries (SFA) were the most commonly affected segments. Of the patients, 39% had claudication while 48.8% had rest pain, and 12.2% of patients had no pain. Trophic changes were seen in 78.5% of the affected limb, 68.3% were chronic smokers, 41.5% had diabetes, and 29.3% had hypertension. Out of the 37 patients who had a recent lipid profile assessment, 62% had low HDL levels and 16% had high levels of either LDL, cholesterol, or triglycerides.
The aortoiliac segments were difficult to assess because of overlying bowel gas and abdominal wall movements. Depending on the segment assessed, sensitivity ranged from 75–100%, specificity 97.8–100%, positive predictive value 85.7–100%, and negative predictive value 96.6–100%. There was perfect agreement in the aorta and excellent agreement in the common iliac and external iliac segments (Table 1, Graphs 1 and 2).
The femoropopliteal region was easier to scan, except for the deeply situated profunda femoris and the distal SFA. The sensitivity ranged between 69–96%, specificity 92–95%, positive predictive value 70–92%, and negative predictive value 88–98.3%. Concordance was excellent in all segments except in the popliteal artery, which showed only fair agreement (Table 1, Graphs 1 and 2).
In the tibioperoneal region, there was only fair agreement between the two modalities, the posterior tibial artery being slightly better. Here, the sensitivity ranged from 44.4–66.7%, specificity 94.3–100%, positive predictive value 75–100%, and negative predictive value 81–89.6% (Table 1, Graphs 1 and 2).
No adverse event occurred during any of the color Doppler studies. A complication during an angiogram. Extravasation occurred while an access sheath was being repositioned during a downhill puncture. This was well controlled with compression.
Many studies have used a similar methodology to compare color Doppler imaging with angiography in PAD2,7–9 (Tables 2 and 3). While some have used conventional angiography as the gold standard, the present study has used DSA. Most of the previous studies did not evaluate the infrapopliteal vessels,2,7,10 while one has evaluated only the origins of these vessels,9 and another did not attempt to differentiate between stenosis and occlusion.8 This study has examined the infrapopliteal arteries, along their whole length and has also tried to differentiate between stenotic and occlusive lesions.
Color Doppler was able to identify the location and extent of stenotic and occlusive lesions, evaluate vessels distal to an occlusion, and also pick up intimal thickening and wall calcification.
False positive occlusions were due to vessels having no demonstrable blood flow on color Doppler, but were well visualized using DSA. This could be due to extremely sluggish blood flow in the vessel, with collaterals diverting away most of the blood. Diffuse proximal disease is known to cause this.11 Other causes are obscuration by overlying bowel gas, respiratory movements, or heavily calcified vessel walls. It is also known that the distal third of the SFA is a difficult area for Doppler to evaluate due to its deep location.7
False negative occlusions are due to segments seen to be patent on color Doppler, but not visualized using DSA. Nonvisualization with DSA could be due to filling with non-opacified blood, especially in a segment distal to an occlusion.12,13 Thus, DSA can overestimate the length of an occluded segment, resulting in a lowering of the sensitivity of color Doppler.14,15 Another reason could be the inadvertent Doppler sampling of a collateral vessel, while the occluded main arterial segment remained unidentified.7,11,16
False positive stenoses are due to vessels seen as stenosed by color Doppler, but as normal on DSA. This is a limitation of uniplanar angiography, as the vessel can appear normal if the plaque is on an anterior or posterior wall.
False negative stenoses are due to a segment reported as stenosed by color Doppler but seen as normal on DSA. This could be due to either lesions missed by uniplanar DSA or mistaken reporting by Doppler, due to its high degree of subjectivity. It could also be influenced by poor visualization.
Earlier studies evaluating color Doppler imaging have shown varying degrees of sensitivity and specificity. We have observed not only a similar trend in the results, but also an improved concordance between the two modalities.2,7–9,17,20
Excellent concordance was seen in the aorta, iliac, and femoral arterial segments, while the popliteal segments showed a lower concordance. The infrapopliteal segments showed only fair concordance, the posterior tibial artery being relatively better. Still, the concordance was higher than that in previous studies, which found Doppler to be inaccurate in the infrapopliteal vessels.9,18 This could be due to the increased incidence of calcified vessel walls obscuring visualization and reducing the sensitivity of the color Doppler. Poor or delayed opacification of these segments on DSA is also a reason for inaccuracy.
Of the segments that were evaluated, 32% had calcified walls. While visualization was possible if the lumen was large and the degree of calcification was less, the segments that showed marked wall calcification were obscured on the Doppler and were not used for comparison.
The distal SFA, though deeply situated, did show a reasonably good agreement in the cases where visualization was possible.
The poor sensitivity of color Doppler in evaluating the popliteal artery could be due to the lack of anatomical demarcation between the distal SFA and the popliteal artery, resulting in a doubt whether distal SFA disease involved the proximal popliteal artery also. Moreover, DSA has a tendency to overestimate the length of occlusion of the distal SFA, resulting in a poor concordance between color Doppler and DSA.
In the infrapopliteal vessels, the anterior tibial artery (ATA) showed a worse result than the (posterior tibial artery (PTA), despite being more superficial. This could be due to poor visualization of the proximal portion as it passes obliquely through the muscles to enter the anterolateral compartment of the leg. The presence of calcification also obscured visualization and affected accuracy.
An attempt was made to answer the research question: “Can color Doppler replace arteriography in the diagnosis and staging of peripheral arterial disease of the lower limbs?”
Color Doppler imaging was found to have a high negative predictive value and could exclude a significant lesion, thus helping to avoid diagnostic arteriography in a mildly symptomatic patient.
Color Doppler scan could determine the nature and extent of arterial disease. Based on these results, treatment can be planned, either endovascular or surgical. While the angiographic approach and technique can be planned beforehand, accuracy can be improved with reduction of examination time and contrast load, if portions of the arterial tree are previously known to be normal. If the size and grade of the lesion is known beforehand, the balloon and stent of correct dimensions can be kept ready.
Color Doppler can be used to document disease progression and to follow up the results of interventions. Early detection of lesions allows timely and effective management.
However, in patients requiring distal femoral revascularization involving the infra-popliteal arteries, diagnostic arteriography is still necessary to plan treatment.
Thus, color Doppler imaging is indispensable in the diagnostic workup of patients with PAD. The results of the present study suggest that color Doppler may be the only imaging test required in the evaluation of PAD above the level of the knee and could be used to plan treatment.
DSA, the gold standard for this study, has its limitations. Magnetic resonance angiography (MRA) has an increased sensitivity and can even demonstrate the patency of segments not seen on conventional arteriography. When compared to intra-arterial pressure measurements, arteriography is only 69% sensitive and 75% specific for the identification of hemodynamically significant (>/= 50%) aortoiliac stenoses.3
There are no anatomical landmarks between adjacent segments, making precise localization of lesions difficult. This was especially noticed between the distal SFA and the popliteal artery.
Better subtraction methods are required to visualize vessels that are poorly opacified with contrast.
Usage of biplanar angiography would have improved the accuracy of the gold standard.
Suggestions for further studies. A more clinically relevant study would be to look at the effectiveness of different modalities in terms of surgical planning and patient outcome. Further studies should focus on the infrapopliteal vessels. Newer technologies, such as computed tomography angiography and MRA, should be compared with color Doppler imaging. If larger numbers of patients are available, the performance of the imaging tests on different subgroups of patients can be assessed, particularly in those are at higher risk of adverse events (diabetes, renal insufficiency).
The study population is representative of the patient load in a hospital setting. A further study is being planned to assess patients in a community setting using color Doppler.