Applied Radiology

Dr. Devane is a third-year resident at the University of Alabama at Birmingham. He completed an internship in internal medicine at UAB and his first year of radiology at the University of Florida. He graduated from the University of Florida College of Medicine and will begin his fellowship training in vascular and interventional radiology at UAB in 2002.

Hypertension is one of the most prevalent diseases in the United States, affecting an estimated 50 to 60 million people. ¹ Renovascular disease is the most common potentially curable cause of hypertension. While only 2% to 5% of hypertension cases have a renovascular cause, this still represents a significant number of patients. The need for a safe, cost-effective, noninvasive screening examination is clear. Three-dimensional contrast-enhanced magnetic resonance angiography (CE-MRA) has evolved into an important diagnostic tool that may fulfill this need.

Causes and treatment options

Renovascular hypertension is caused by an arterial stenosis that incites the renin-angiotensin system to increase glomerular filtration. The angiotensin II that helps improve glomerular vasoconstriction also causes peripheral vasoconstriction and thus systemic hypertension. History and signs suggesting renovascular disease include hypertension refractory to multiple medications, new onset hypertension before age 25 or after age 55, or an abdominal bruit. ² The two most common causes of renovascular hypertension are atherosclerosis and fibromuscular dysplasia.

Atherosclerosis is the leading cause of renal artery stenosis, primarily affecting the ostium, the proximal renal artery, or both. This disease typically affects older patients. Fibromuscular dysplasia affects a younger population, usually women between the ages of 30 and 40 years. Unlike atherosclerosis, fibromuscular dysplasia usually affects the mid and distal renal arteries.

Both forms of renal artery stenosis can be treated with percutaneous transluminal angioplasty (PTA), without or with stenting, often with varying degrees of success. Fibromuscular dysplasia is generally considered more amenable to angioplasty than atherosclerotic disease. Tegtmeyer et al ³ reported a 39% cure rate and 59% rate of symptom improvement for patients with fibromuscular dysplasia treated with PTA. Surgical revascularization such as splenorenal or aortorenal bypass are therapeutic options for disease not amenable to angioplasty. ¹ Accurate determination of disease presence and extent prior to anticipated therapy can expediently direct patients to the preferred mode of treatment, and can help avoid unnecessary additional tests and procedures with their associated risks.

Diagnosis of renal artery stenosis

Conventional catheter angiography has historically been the gold standard diagnostic test for renal artery stenosis. Unfortunately, this test is expensive, invasive, and requires nephrotoxic contrast material. Angiography can also overestimate or underestimate disease because lesions cannot be viewed from more than one or two perspectives. Furthermore, conventional an-giography does not yield information about the physiologic significance of visibly narrowed arteries; not all apparent stenoses produce hypertension.

The most commonly performed noninvasive functional test has been captopril renal scintigraphy. The sensitivity and specificity of this procedure have been reported to be 80% and 100%, respectively. ⁴ Captopril scintigraphy is less effective in patients with bilateral disease or significant renal insufficiency. Scintigraphy also does not provide anatomic information that can be crucial in planning treatment.

Some institutions use ultrasound to detect renal artery stenosis, with variable and often limited success. Ultrasound is highly operator dependent and is a technically difficult exam, even with expert operators. Also, standardized methods and criteria for stenoses have not achieved widespread acceptance. ⁵

CT angiography (CTA) is a promising emerging modality, particularly with the advent of multidetector technology. CTA is noninvasive and provides three dimensional information. Unfortunately, like conventional an-giography, CTA exposes the patient to radiation and to the risks of iodinated contrast, such as nephrotoxicity and possible allergic reactions.

The various problems associated with all of the above modalities make MRA an intrinsically appealing alternative, if it can provide accurate information in a cost-competitive manner.

Magnetic resonance angiography

Prior to the development of 3D CE-MRA, the most commonly used methods of MRA were non-contrast time- of-flight and phase contrast. Time-of-flight MRA utilizes short TR (repetition time), short TE (echo time), partial flip angle gradient echo techniques to saturate signal from stationary tissue. Blood flowing into the plane or volume of imaging is unsaturated and consequently produces a much brighter signal. Flow signal from multiple different slices can then be combined to create an image resembling an angiogram. Though each individual slice can be acquired quickly, complete studies can be time consuming.

Phase-contrast MRA works by subtracting background tissue from the signal from flowing blood. ⁶ An accurate estimation of the velocity within the renal artery is required, which is often difficult to accomplish. ⁷ It is also a time-consuming sequence, usually taking even longer than time-of-flight techniques covering comparable territory.

The limitations of non-contrast time-of-flight and phase-contrast techniques include artifact from breathing or other patient movements, limited contrast to noise resolution, signal loss from turbulence, and in-plane flow saturation. ⁸ Because of the oblique orientation of the renal arteries relative to the aorta, and because of motion from breathing, renal MRAs using these techniques are often disappointing. Only the proximal renal arteries can be reliably imaged, and successfully identifying accessory renal arteries can also be difficult (figure 1).

3D contrast-enhanced magnetic resonance angiography

The advent of systems with improved gradient strengths and software makes dynamic, contrast-enhanced MRA possible. This is a 3D spoiled gradient echo sequence which covers the entire territory of interest in a single breath hold immediately following intravenous (IV) administration of a gadolinium chelate contrast material. Compared to non-contrast techniques, this method is very fast, gives improved contrast-to-noise resolution, and virtually eliminates motion artifacts in cooperative patients. Signal loss from in-plane flow is eliminated, and signal loss from turbulence is substantially reduced. The renal arteries have proven ideal targets for this new technique. Multiple studies have shown 3D CE-MRA to be both sensitive and specific in detecting renal artery stenosis (Table). ⁹

Gadolinium-based contrast agents solve many problems that plague non-contrast MRA. Gadolinium causes marked T1 shortening, providing a bright signal. The short TE/TR 3D spoiled gradient echo sequence is very sensitive to these T1 shortening effects. If the sequence is timed so acquisition occurs when the contrast bolus is highly concentrated in the arteries, the result is a set of images in which the arteries are the only bright objects. The images can then be combined and reformatted in any plane, giving a truly three-dimensional angiogram.

Gadolinium agents' advantages over iodinated contrast include lack of significant nephrotoxicity in standard doses and extremely low incidence of allergic reactions. ¹⁰ The lack of nephrotoxicity is important with the high incidence of concomitant renal insufficiency in patients with renal artery stenosis. Like iodinated contrast, IV gadolinium chelates remain exclusively in the arteries for only a short period of time. To be most effective, the MR sequence must be completed during a brief interval between arrival of contrast in the target arteries and the appearance of contrast in the nearby veins. This time frame is usually under 30 seconds, thus requiring careful coordination between contrast administration and scanning.

Technique specifics

The basic sequence for 3D contrast-enhanced renal MRA is a three dimensional spoiled gradient echo sequence with minimum TR and TE, and medium-to-high flip angle (45 to 60 degrees). The imaging volume is oriented in a coronal or coronal-oblique plane, reaching from just above the celiac axis to the groin, with slices under 3 mm thick. On most modern high field magnets, this sequence can be completed in under 30 seconds. Either a body coil or phased array torso coil can be utilized. The patient is placed in the supine position with arms elevated above the head to avoid phase wrap artifact. Images are acquired during a single breath hold to limit breathing artifacts that can severely degrade the images. ¹¹ Supplemental oxygen may be helpful to increase breath hold time.

IV contrast consists of 0.1 to 0.2 mmol/kg gadolinium (some institutions prefer to give every patient two bottles or 40 mL, which is equivalent to 0.2 mmol/kg in many patients). The contrast can be given with either hand or machine injection, though many institutions prefer power injectors to improve consistency. The timing of the bolus is crucial; there is not much margin for error as venous enhancement occurs just seconds after the arterial bolus reaches the kidneys, contaminating the images.

It is also important to understand the order in which data are acquired during the study, in order to be sure that the key portions of the sequence occur during optimal arterial enhancement. "k-space" refers to the Fourier data acquired during the imaging sequence. The center of k-space encompasses the low frequency data which provides most of the image contrast, while the periphery of k-space is the high-frequency data mostly responsible for image detail. In MRA, it is important to acquire the center of k-space during peak arterial enhancement. Two commonly utilized methods of forming k-space data are sequential and elliptical-centric. The elliptical-centric method samples near the center of k-space first, providing the best chance of capturing only the arterial signal, if the bolus of contrast was properly timed (figure 2). ¹² Machines limited to sequential filling of k-space can still perform high quality studies, though correct timing can be more complicated.

There are several commonly used methods for aligning the contrast bolus with the acquisition of the center of k-space:

* Making a "best guess" estimate, which in most patients is approximately 15 to 20 seconds. This is the easiest but runs the greatest risk of being either too early or too late, which cause artifacts that can complicate interpretation (figure 3).

* Giving a 1 to 2 mL test bolus of gadolinium while doing sequential images at a fixed level in order to time the flow of contrast from the injection site to the aorta. ¹³ The best time for beginning the sequence is then calculated depending on the method chosen for filling k-space. An additional minor limitation of this method is slight venous and parenchymal enhancement can contaminate the subsequent arterial images.

* Using an automated bolus tracking technique (such as SmartPrep [GE Medical Systems, Waukesha, WI] or Bolus Trak [Philips Imaging Systems, Shelton, CT]) to detect contrast bolus arrival in the aorta and trigger the scan.

* Performing serial preliminary scans at the level of the renal arteries during injection of contrast and manually triggering the actual MRA sequence when arterial enhancement occurs; this method is commonly referred to as "MR fluoroscopy" (figure 2). ¹²

* Repeatedly performing a complete but very rapid MRA sequence during and after contrast injection, analogous to conventional digital subtraction angiography (DSA). This eliminates the need for bolus timing. ¹⁴

Additional features that can improve a CE-MRA study include partial Fourier (or partial NEX) data sampling, ¹⁵ rectangular field of view, and zero fill interpolation. All of these features can improve spatial resolution, temporal resolution, or both, depending on what one chooses.

Post processing and image review --Post processing and interactive review of images is a critical component of CE-MRA. This requires using either a free-standing workstation or the operating console of the MR scanner to view the data from multiple perspectives and ultimately record images that will be accurate and convincing for both the interpreting radiologist and the referring clinician. Because a correctly timed renal MRA has little or no signal from background tissue, there is generally no need for time-consuming manual excision of surrounding structures from the images. Total post-processing time is generally less than 5 to 10 minutes.

Multiplanar reformatting is the simplest method for viewing the data. Many workstations do this automatically, allowing the radiologist to view slices of the data in coronal, axial, sagittal, or oblique planes. The thickness of the reformats can be chosen by the operator, with thin slices often being most useful in image review. Axial reformats, in particular, are useful in identifying accessory renal arteries and unusual branching patterns.

The most common way to record the images for final display is with a maximum intensity projection (MIP) technique. An MIP image is a two-dimensional image that maps the brightest pixels in the data set from a given perspective looking through the entire data volume. Darker pixels in front of or behind the arteries are not included in the MIP display. Background tissue, which is mostly dark pixels, will also not interfere. The result is an image that looks like a conventional angiogram. Unlike a conventional angiogram, however, the data are three-dimensional; a MIP image can be rotated to view the vessels from any perspective.

Subvolume MIP images can be constructed from a limited volume within the original full volume of data. This subvolume MIP can be obtained in any desired plane to exclude overlapping arteries and venous contamination. This technique is particularly useful for renal MRA, as signal from the mesenteric, splenic, hepatic, and lumbar arteries can obscure the renal arteries on full-volume MIP displays. Subvolume MIPs can also help deal with venous contamination by locating the volume behind the renal veins. Subvolume MIP images are often especially effective at conveying the study's information in a way that referring clinicians can readily perceive and accept. The radiologist or technologist performing the post processing should always be sure to record and film the images that accurately reflect what the radiologist reports in his or her interpretation (figures 4 and 5).

MRA data can also be displayed in a surface-rendering format that can be used to analyze the internal or external surface of the vessel wall. These displays are limited by their inability to provide accurate information about the degree of stenosis, and are not usually used for routing image interpretation. Workstations with more advanced volume rendering capabilities can also be used for MRA interpretation, but since CE-MRA does not have much signal from nonvascular structures, there is little benefit.

Advantage and disadvantages

Three-dimensional CE-MRA is noninvasive, nonnephrotoxic, provides images in three dimensions, and doesn't expose patients to ionizing radiation. The 3D capability is a clear diagnostic advantage over catheter angiography. In fact, many investigators believe that in some instances MRA is just as sensitive, if not more sensitive, than catheter angiography because of the multiplanar capability. ¹⁶ Overlapping vessels can be eliminated to demonstrate accessory vessels and eccentric stenoses that could not be detected by catheter angiography.

The noninvasive and safe nature of MRA appeals to both patients and referring clinicians. Patients also appreciate that this study can easily be completed within a routine 30 to 45 minute time slot, avoiding the need for a night in the hospital, an observation period, or even time away from work. Consequently, MRA may promote increased referrals, leading to increased or earlier detection of disease and, it is hoped, improved treatment of renal artery stenosis (figures 6 and 7).

Although 3D CE-MRA provides accurate anatomic information about renal arteries, the technique described above fails to provide physiologic information about the significance of any detected disease. Like catheter angiography, and unlike captopril renal scintigraphy, MRA is an ana-tomic study. Some investigators favor using phase-contrast MRA as an adjunct to CE-MRA in order to gain more physiologic information. The presence of signal loss due to dephasing on a phase-contrast study suggests that the stenosis is flow limiting. ¹⁷ The value of performing additional phase-contrast MRA is controversial. It still does not prove that the patient has renovascular hypertension, and its adds a significant amount of time to the examination.

Some institutions perform captopril scintigraphy in addition to MRA. Other institutions obtain only CE-MRA without a phase-contrast study and either treat the patient based on anatomic information alone or in combination with data from captopril scintigraphy. An argument can be made that if two tests will always be necessary, then catheter angiography should be pursued in all patients with a positive captopril scintigram, making MRA unnecessary. The weakness of this argument lies with the intrinsic weaknesses of the captopril study, discussed above. Nevertheless, the limited physiologic information provided by current MRA techniques leaves room for future improvements, such as the possibility of combined MRA-captopril studies. ¹⁸

Currently the spatial resolution of DSA and CTA are superior to that of MRA (1 mm ³ ). Many investigators feel that the spatial resolution of MRA is insufficient to evaluate for fibromuscular dysplasia and stenoses in small branches and accessory vessels. Young patients with a higher likelihood of having fibromuscular dysplasia may thus be better served by catheter angiography, given current MR capabilities. However, continued improvements in MR technology may settle the issue in MRA's favor in the near future.

CE-MRA also requires skill, experience, and patience on the part of both radiologists and technologists. The dynamic nature of the examination, the precise timing required, and the differing capabilities of different equipment necessitate attention to detail and good communication among the MR imaging team. Some practice is necessary before excellent images can be consistently obtained. The equipment is expensive, with study working best on state-of-the-art, high field magnets. Additional technologist training may be necessary to perform the study reliably. Some radiologists prefer to monitor the study in person and perform the post processing themselves.

At our institution, we perform the study in about 30 minutes with reformats and image interpretation requiring about 10 more minutes. If the radiologist supervises the whole exam, this time commitment may be comparable to performing a catheter angiogram. However, once technologists become accustomed to the technique, radiologist involvement during the scan can shrink substantially. Similarly, experience with post processing can reduce the time required for that component and technologists can be trained to perform most of the post processing and filming.

The future of angiography

Some radiologists speculate on whether MRA will completely replace catheter angiography, not only for renal arterial imaging, but for all vascular imaging, including coronary angiography. Limitations of spatial and temporal resolution with MRA make this an unrealistic goal in the near future, though MRA may soon replace diagnostic angiography for many indications. Renal and carotid artery studies, and perhaps lower extremity runoff examinations, are currently better suited for MRA than other types of vascular studies, such as coronary imaging.

Much debate has arisen as to how MRA will impact the field of vascular-interventional radiology. Many an-giographers feel it will threaten their livelihood. It is clear that MRA will alter the nature of the angiographer's business, though not necessarily in an adverse fashion. Vascular-interventional radiologists can keep themselves involved with diagnostic angiography by becoming experienced with MRA. MRA may also identify more patients who could benefit from percutaneous intervention, in addition to noninvasively determining who will not benefit. Thus, the interventional radiologist may see an increase in numbers of patients referred for therapy, while the number of negative diagnostic angiograms should drop. Both the patient and the radiologist will thus benefit.

Conclusion

Three-dimensional breath-hold contrast-enhanced MRA is a safe, noninvasive, and sensitive test for renal artery stenosis. When utilized in patients with a high clinical likelihood of renovascular hypertension, CE-MRA can provide highly diagnostic images, leading the way to treatment by angioplasty or surgical revascularization. Current limitations in spatial resolution reduce MRA's value in diagnosing fibromuscular dysplasia and peripheral stenosis. However, in older patients more likely to have atherosclerotic disease, CE-MRA is a highly effective tool. Patients with renal insufficiency or allergy to iodinated contrast are ideal candidates for CE-MRA. With continuing technologic improvements and development of techniques providing more physiologic information, MRA is likely to become the test of choice for most patients with suspected renovascular hypertension.

Acknowledgements

The author would like to thank Dr. Brian Jones for mentoring this project and Dr. J. Kevin Smith for editorial assistance, Anthony Zagar for photography, and Pat Moore for technical assistance.