With the advent of color encoding of the Doppler signal, US has gained a major place in the detection of renal artery stenosis. To be of diagnostic use, a complete examination must be undertaken, including B-mode, spectral sampling, and color imaging.#

The stenosis may be observed on color-flow images with focal changes of color (Fig. 1A) and/or a perivascular artifact, which are related to acceleration and turbulence, respectively. Proximal criteria of hemodynamically significant stenosis essentially are based on spectral sampling showing spectral broadening and increased velocity (Fig. 1B). The main proximal criteria used are first, a renoaortic velocity ratio greater than 3.5 for a 60% stenosis⁴⁵ and second, a peak systolic velocity at the site of stenosis, with an upper limit of either 150 cm/s for a 50% stenosis⁶ or 180 cm/s for a 60% stenosis.⁷ Using these criteria, the sensitivity and specificity of the technique compared with angiography are 89% to 98% and 90% to 98%, respectively.^7-16 Although multiple renal arteries can be detected, in all series, the detection of small accessory arteries has been reported as very low.#

Severe stenoses that decrease the diameter by >75% produce distal intrarenal spectral changes, a tardus-parvus phenomenon with a slowed systolic acceleration and a decreased resistive index¹⁷ (Fig 1c). Many distal quantitative criteria have been proposed in the literature (loss of early systolic peak; acceleration lower than 3 m/s²; acceleration index >4, acceleration time of systolic peak >0.07 s; a difference between kidneys in resistive index of >5% or in pulsatility index of >0.12). The performance of correlative studies with angiography is confusing because of the variability in criteria and in corresponding degree of stenosis.^14,16,17-21 Interobserver and intraobserver variability using these criteria is high.²²²³ The factors responsible for changes in the distal waveforms are complex and are probably more dependent on changes of compliance than on pressure decrease.²⁴ Therefore, these criteria are used only when obvious on spectral traces, when quantifying the stenosis as severe (>75%), or when identifying a downstream pattern of a stenosis on a segmental or an accessory artery that has been missed.#

The use of slip-ring CT scanners capable of contiguous tube rotations coupled with continuous table incrementation allows for the volumetric acquisition of imaging data from the abdominal aorta, including renal arteries, during a single breath hold. With single-ring CT scanners, Rubin and coworkers²⁶ showed that maximum intensity projections provided better sensitivity (92%) than surface shaded display (59%). However, Johnson and coworkers²⁷ reported that the volume rendering technique provided a better specificity (99%) than maximum intensity projection (87%). Performance of CTA in detecting significant stenoses is good with high sensitivity (88-96%), high specificity (83-99%),^28-30 and close correlation with angiography for grading stenoses.³⁰ The detection of accessory arteries is much better than with Doppler US (average 90%). CT also shows the exact nature of stenoses, eg, ostial, truncal or pseudotruncal³¹ and the degree of calcification of lesions before endovascular treatment.#

With multislice CT, which allows between 4 and 16 slices per rotation, the axial resolution is improved, approaching 1-mm slice thickness (Fig. 2), and the amount of iodinated contrast agent can be reduced. Because of a better spatial resolution, the separation of calcification and true lumen is more accurate,³² and distal lesions, as fibromuscular dysplasia, are better delineated. Preliminary results of this technique are extremely encouraging, but comparative studies with angiography are still not available.#

MRA has now moved from flow-enhanced (time-of-flight or phase-contrast) sequences to T1-weighted contrast-enhanced acquisitions.³³ Its performance is excellent, with a sensitivity and specificity for diagnosis of significant stenosis between 88% and 100% and between 71% and 99%, respectively.^34-42 This technique shows the entire course of renal arteries up to the renal sinus in most cases and the complete abdominal aorta, including its bifurcation (Fig. 3). Most accessory arteries are therefore shown. The evaluation of the degree of stenosis with this method has the same interobserver variability as conventional angiography.⁴³ Technical improvements allow shorter acquisition times and/or higher spatial resolution.⁴⁴⁴⁵#

Comparison of performance of these noninvasive tests is difficult. A recent meta-analysis⁴⁶ tried to compare the validity of CTA, MRA, and US for diagnosis of RAS in patients suspected of having RVH. Receiver-operating characteristic curves found that CTA and gadolinium (Gd)-enhanced 3-dimensional (3D) MRA performed significantly better than the other diagnostic tests and seemed to be preferred in patients referred for evaluation of RVH. However, because few studies of these tests have been published, further research is recommended.#

When the stenosis is severe, a poststenotic dilation occurs as a “jet-lesion.” This dilation can be used as a criterion of significant artery stenosis. This morphological change can be assessed with CTA and MRA but not with US. However, this criterion is difficult to quantify; no significant threshold has been defined, even if a 20% dilation is widely used; it has never been evaluated to our knowledge.#

When the renal blood flow is significantly decreased, renal parenchyma shrinks. Several parameters have been proposed to evaluate this effect. Renal length can be measured with any imaging technique. To be significant, a length difference of 1 cm should be considered, attesting of a hemodynamically significant stenosis.⁴⁷ If the renal length is less than 8 cm, revascularization is contraindicated because less likely to be of benefit.#

Measurement of cortical thickness and cortical area has been proposed by Mounier-Vehier and coworkers,⁴⁸ who showed that thresholds of 8 mm for cortical thickness and of 800 mm² for cortical area allowed the differentiation of control kidneys from poststenotic kidneys, whereas renal length was still within normal range, suggesting that cortical parameters are more sensitive for early diagnosis of atherosclerotic renal disease than kidney size (Fig. 4). Cortical atrophy seems to be a useful marker for guidance in revascularization, but its prognosis value has still to be evaluated.#

The impact of RAS on glomerular filtration and renal perfusion has been studied with MRI.#

Semiquantitative evaluation, as split renal function, usually is sufficient in urological management of most uropathies. Rohrschneider and coworkers⁶⁶ obtained calculations of the percentage of the single-kidney “activity” comparable with those derived with gamma camera scintigraphy.⁶⁷ These studies were based on a dynamic gradient-echo sequence and half of a standard clinical accepted dose of Gd-DTPA. A region of interest was positioned around the renal parenchyma (omitting medulla and pelvis), and calculation of the relative renal function was then based on the equation: RF=AUC⁢ (mm2)×S⁢ (mm2) #

where AUC corresponds to the area under the a part of the time-intensity curve (Fig. 7) and S is the region of interest. The split renal function (in percentage) corresponds, for each kidney, to the product: RF(%) = RF/RFtotal × 100, where RFtotal is the sum of RFs of both kidneys.#

On the behalf of these comparable results with scintigraphy and because it is able to provide morphological and functional evaluation of the urinary tract during the same imaging session, it is likely that MRI may be used more systematically in a near future.#

On the contrary, measurement of GFR is used as an index of functioning renal mass, representing the sum of filtration rates in each functioning nephron. A decrease in GFR may be the earliest and only clinical sign of renal disease, and its serial monitoring allows one to estimate the severity and to follow-up the course of kidney diseases. The measurement of GFR using MRI is a real challenge, and attempts have been unsuccessful until now. Several studies, mostly performed in animals, using different types of models showed correlation with scintigraphic techniques but with variable overestimation or underestimation.^68-71#

An extensive review of all technical problems related to quantitative GFR measurement has been published recently.⁷² It requires sampling of abdominal aorta, with both kidneys, with a sufficient time resolution to accurately define the arterial input function and to separate adequately the cortical vascular phase (perfusion) and the filtration phase. Arterial signal intensity-time curve is used in different kinetic models to compensate for the noninstantaneous bolus injected into the blood. It also requires acquisition of the information within the whole kidney (and not one slice only) using fast 3D sequences. T2* effects of the contrast agent during its maximal concentration phases (intra-arterial bolus and intramedullary water reabsorption) must be minimized by a decreased injected dose and a proper selection of the parameters of the imaging sequence.#

Another major problem is the necessity, for accurate quantification of these physiological parameters, to convert SI values into concentrations. Several methods, all with significant limitations, have been proposed, including scanning a phantom of tubes filled with Gd solutions at various concentrations, with the sequence used for the dynamic study, to obtain a calibration curve and a conversion equation. The solution could be to measure dynamically the R1 relaxivity instead of measuring signal intensity (based on Look-Locker sequences) because R1 is linearly correlated to tracer concentration.⁷³ Such approaches are still under development and validation. Finally, adequate postprocessing techniques must be used to correct for respiratory movements of the kidneys and automatic measurement of the cortical volume. If all of these requirements are overcome, and when validation is obtained with adequate goldstandards, dynamic MR will be able to provide single kidney GFR values as well as intrarenal GFR maps (Fig. 8).#

These quantitative methods are still not ready to use, and many technical problems remain challenging. Therefore, scintigraphy still remains the gold standard in that field.#

Recently, a quantitative method of in vivo measurement of the single kidney EF was proposed by Dumoulin and coworkers⁷⁴ and implemented by Niendorf and coworkers.⁷⁵ This method is based on the measurement of T1 within flowing arterial and venous blood during slow Gd-infusion. Once the EF is calculated for each kidney, the GFR also can be calculated by measuring RBF of each renal artery using the cine-phase-contrast method. Preliminary results on animals have shown concordant results with the reference method (inulin clearance). If this quantitative method could be transposed to humans with reproducible results, the applications in nephrology would be numerous.#

Outer medulla is particularly sensitive to hypoxia because the active reabsorption process within the thick ascending loop of Henle requires high level of oxygen consumption.⁷⁶ Therefore, a decrease of medullary blood flow, as in acute renal failure, or an increase in tubular reabsorption, as in diabetic nephropathy (at the stage of hyperfiltration), may induce medullary hypoxia and secondary ischemia.#

The blood oxygen level-dependent (ie, BOLD) technique does not measure directly pO2 but allows intrarenal R2* (1/T2*) measurements, which are closely related to concentration in deoxyhemoglobin (Fig. 9).⁷⁷⁷⁸ Therefore, absolute R2* values cannot be used in practice. If the disease is asymmetric, as in renal artery stenosis, static comparison of both kidneys may identify hypoxia on one side. In renal parenchymal diseases, only dynamic changes after physiological or pharmacological manipulation can identify the kidney response.#

Interestingly, diabetic subjects are unable to increase the medullary oxygenation during water diuresis. In a comparative study with a matched control group, diabetic patients without microalbuminuria, hypertension, or renal insufficiency did not show any significant improvement of medullary oxygenation after water load.⁷⁹ Wider applications of this technique to clinical ischemic conditions of the medulla still have to be established.#

Macrophages, virtually absent in normal kidneys, may infiltrate renal tissues in specific nephropathies such as acute proliferative types of human and experimental glomerulonephritides, renal graft dysfunction (rejection and acute tubular necrosis), and in acute ischemic disease.⁸⁰ Their role is complex, contributing to glomerular and tubulo-interstitial injury through the secretion of various cytokines and proteases which induce changes in extra-cellular matrix and progressive fibrotic changes (glomerulosclerosis, tubulointerstitial fibrosis).#

Ultra-small superparamagnetic particles of iron oxide (ie, USPIO) are small-sized nanoparticles that have a long half-life in the bloodstream (2 h in rats and 36 h in humans) and are avidly captured several hours after intravenous injection by extrahepatic cells with phagocytic activity which include blood circulating monocytes and resident macrophages that are present in most tissues.#

Several models of experimental nephropathies in rats were used to demonstrate the detectability of intrarenal macrophagic activity in vivo (Fig. 10).^81-84 The degree of decrease of SI was always correlated with the number of macrophages within each renal compartment and the severity of disease. The results of the first pilot clinical study⁸⁵ seem to corroborate experimental results and justify larger multi centers clinical trials.#

In acute renal failure models, the primary site of renal injury (proximal or distal) remains uncertain. Some authors could demonstrate proximal tubule dysfunction in acute tubular damage induced by cysplatin in mice⁸⁶ using Gd-dendrimer chelates, which are cleared by glomerular filtration.#

In acute renal failure, direct toxicity to the cells or ischemic damage may lead to apoptosis or necrosis of tubular cells. Therefore, the identification of these cellular events could help in assessing the severity and the reversibility of diseases.#

SI observed on diffusion-weighted sequences depends on 2 major factors: First, on molecular movements within intracellular and extracellular spaces which depend essentially on local temperature and biological barriers present in the tissue (cellular membranes, fibers, organelles, macromolecules, etc.), and second, on “external” factors, such as perfusion, magnetic susceptibility of tissues, cardiac pulsations and respiratory movements.⁸⁷#

Using an appropriate acquisition method, cortical values of apparent diffusion coefficient (ADC) are greater in the cortex than in the medulla, and medullary diffusion is anisotropic (spatially oriented).⁸⁸ In theory, diffusion-weighted imaging could be useful to separate cellular edema, ie, reversible damage with decreased ADC values, from cellular necrosis, ie, irreversible damage with increased ADC values. In an experimental model of diabetic nephropathy, we showed that the regional ADC values were decreased within the outer medulla, when tubular cell edema occurred.⁸⁹ However, the exact role of this method in evaluating prognosis of acute renal diseases has still to be defined.#