Introduction

End-stage renal disease (ESRD), defined as complete or near complete renal failure, is an increasingly important medical problem, ultimately requiring hemodialysis in the vast majority of patients. The incidence and prevalence of ESRD as well as the number of patients requiring hemodialysis have sharply risen over the past few years. In 2004, an estimated 1.22-million patients were on hemodialysis world-wide, representing a 20% increase in 3 years since 2001 [1, 2].

A well-functioning vascular access is the cornerstone of hemodialysis treatment and can be achieved by insertion of a central venous catheter or by surgical creation of an arteriovenous fistula or graft (Fig. 1). The access of first choice is the autogenous arteriovenous fistula (AVF) because of its better long-term performance and patency rates when compared with arteriovenous grafts (AVG) and central venous catheters (CVC). Furthermore, AVFs have lower vascular access related morbidity, mortality and healthcare costs compared with AVG and CVC [3, 4]. In contrast, prosthetic vascular grafts require about five-times more therapeutic interventions compared with AVF to keep the access functioning [5–7].

Fig. 1
figure 1

Schematic overview of a radial-cephalic arteriovenous wrist fistula (left) and a forearm arteriovenous loop graft (right). The graft anastomosed to the brachial artery and cephalic vein at the antecubital crease in an end (graft) to side (vessel) fashion. The loop has been tunneled underneath the skin to enable easy cannulation

Full size image

In order to reduce long-term vascular access related complications, the Dialysis Outcome Quality Initiative (K-DOQI) and Good Nephrological Practice guidelines advocate an all AVF policy, i.e. at least 70% of all newly created accesses should consist of autogenous AVF [3, 4]. However, the major drawback of AVF creation is the relatively high frequency of early thrombosis—up to 10–20% of all newly created vascular accesses thrombose within the first week after creation—and non-maturation [8, 9]. Non-maturation is defined as an AVF being inadequate for hemodialysis due to insufficient flow-volume or insufficient venous distension after creation. Causes of non-maturation are thought to be the use of tiny vessels in addition to arterial inflow or venous outflow stenoses or occlusions [9–13]. Furthermore, the presence of large caliber side branches may also jeopardize AVF maturation due to a disadvantageous flow distribution [3, 14, 15]. Different studies have reported AVF non-maturation rates within the first months after creation from 5% up to 54% [12, 16–20].

In order to increase the number of mature and functional AVFs, adequate history taking, physical examination and preoperative assessment of upper extremity vessels is important [3, 4, 19]. Increasingly, arterial and venous diameters as well as the presence and location of preexisting atherosclerotic occlusive disease and venous stenoses, occlusions and side-branches are used to guide the choice of fistula type and location. Consequently, interest has risen in preoperative imaging of upper extremity vessels. The goal of preoperative imaging is assessment of vessel caliber and identification of sites where arteries and veins are of suboptimal quality for access purposes.

The current review will provide the radiologist with an overview of the clinical role and relative merits and shortcomings of physical examination, duplex ultrasonography (DUS), digital subtraction angiography (DSA) and contrast-enhanced magnetic resonance angiography (CE-MRA) in the preoperative work-up of patients awaiting surgical creation of a vascular access for hemodialysis.

History taking and physical examination

The cornerstone of any pre-surgical workup is a thorough history and physical examination. Negroid race, female gender, diabetes mellitus, peripheral arterial occlusive disease, previous vascular access procedures and axillary radiation therapy are associated with poor vascular access outcome [4].

Careful physical examination is important in the work-up of patients awaiting vascular access surgery. Skin lesions, local infections, generalized dermatological problems and scars, in addition to small arterial caliber and lack of pulse vigorousness as well as small venous caliber and distensibility, may indicate poor chance of successful AVF creation at standard locations and should therefore be documented and addressed [4]. However, arterial and venous assessment by physical examination can be challenging and of limited value in obese patients [12, 21]. Malovrh et al. [9] found that physical examination failed to identify suitable vessels for AVF creation in over half of all patients undergoing dialysis access surgery (n = 62/116; 54%), underscoring the need for additional information prior to access creation.

Duplex ultrasonography

DUS enables assessment of vessel patency, diameter, flow-volume and velocities. The application of DUS enables better depiction of adequate vessels for AVF creation that may not be detected by physical examination, especially in obese patients [9, 12, 21, 22]. Multiple studies have found that the application of DUS resulted in changes in surgical procedure in 31% of cases, changes in site of exploration (9.6%), a decrease in unsuccessful explorations (11% to 0%), an increase in the relative number of AVFs created (as opposed to other types of access; 64% vs 34%), and a decrease in non-maturation rates (38% to 8.3% and 66% to 46%), when compared with the use of physical examination alone [12, 16, 23–25]. However, reported cut-off values of various DUS-derived parameters (i.e. diameter, flow-volume, velocities, compliance and resistive index) are inconsistent. An aggressive approach to conform to the all AVF policy may, therefore, still result in increased early failure and non-maturation rates if based on a single DUS-derived parameter [26, 27]. It remains to be established if a combination of parameters might enable better prediction of vascular access function and minimize early failure and non-maturation rates.

Arterial assessment

Preoperative DUS examination should include assessment of the arteries from the subclavian artery down to the radial and ulnar arteries at the wrist [4]. The exact course and continuity as well as the presence of stenoses should be addressed because patients with arterial stenosis are thought to be at increased risk for developing hand and finger ischemia after AVF creation due to steal phenomena [4]. For detection of relevant stenoses (defined as >50% luminal reduction) in the upper extremity arterial system, DUS has a sensitivity and specificity of 90.9% and 100% for the subclavian artery, 93.3% and 100% for upper arm arteries, 88.6% and 98.7% for forearm arteries and for the arteries of the hand 70% and 100%, respectively [28, 29].

Another important morphological parameter apart from the presence of arterial stenosis is arterial diameter. Arteries with diameters smaller than 1.5 to 3.0 mm have been associated with increased non-maturation rates AVF [11, 24, 25, 30–33].

Additional parameters such as radial artery flow volume and peak systolic velocities before or during reactive hyperemia have also been reported to be predictors of AVF maturation. In Fig. 2, radial artery flow velocity changes due to fist clenching and reactive hyperemia are shown. Lockhart et al. [27], in contrast, found that arterial diameters, resistance indices and peak systolic velocities had only little if any predictive value for AVF outcome.

Fig. 2
figure 2

Duplex ultrasound assessment of radial artery flows at rest (a), during fist clenching (b) and during hyperemia after the fist clench has been released in a healthy volunteer. The arrow indicates the moment of fist clench release. Absence or diminished change in radial artery flow is associated with a higher risk of vascular access early failure and non-maturation

Full size image

Venous assessment

The superficial venous system of the upper extremity is easily assessable by DUS and results in detection of more veins compared with physical examination alone [12, 21, 22]. Furthermore, it also allows for assessment of local hemodynamics, such as subclavian vein flow. A typical example of Doppler signal changes of the subclavian vein during deep inspiration in a healthy volunteer is shown in Fig. 3. The absence of changes in venous Doppler signal due to deep inspiration or loss of venous compressibility is another important finding indicative for local venous stenosis or occlusion. Preoperative detection of stenoses and obstructions is important to avoid unsuccessful surgical explorations [4]. Nack et al. [34] reported a DUS sensitivity, specificity, positive predictive value and negative predictive value of 81%, 90%, 90% and 78% respectively, for detection of venous stenosis, thrombi and occlusions when compared with DSA.

Fig. 3
figure 3

The effect of deep inspiration on subclavian vein flow assessed with duplex ultrasound. The arrow indicates the start of deep inspiration. Absence or diminished subclavian vein flow or change flow due to deep inspiration is indicated for central venous stenosis or occlusion

Full size image

The clinical value of upper extremity DUS for detection of arterial or venous abnormalities is lower for proximal compared with distal arteries or veins, respectively. For instance, Nack et al. [34] reported progressive decreasing DUS sensitivities for detection of abnormalities in the subclavian vein (79%), innominate vein (75%), and superior vena cava (33%), when compared with DSA. This can be explained by the fact that veins course underneath bony structures such as the clavicle and ribs over a substantial length and because the central thoracic veins are relatively inaccessible by DUS because of the distance to the transducer.

As is the case for arteries, DUS-derived venous diameter is an important parameter for prediction of vascular access outcome. For assessment of venous diameter, a proximally applied cuff should be used to induce venous dilatation for better appreciation of ‘maximum’ or ‘true’ venous diameter [3, 4]. Reported venous cut-off diameters for successful AVF creation range from 1.6 to 2.6 mm [9, 19, 24, 25, 30, 33, 35]. This range may be partially explained by differences in vein mapping protocols because only few authors reported the measurement conditions and methods to achieve venous dilatation. Furthermore, DUS venous diameter measurements are observer-dependent with an inter-observer variation of 0.5 mm [36]. Recently, Planken et al. [37] have demonstrated that superficial forearm vein diameter measurements vary over time with a coefficient of variation of 27%. In addition, forearm superficial venous diameter measurement reproducibility depends on the applied venous congestion pressure and best reproducibility is achieved at venous congestion pressures >40 mmHg [38].

Preoperative contiguous length of non-diseased vein >10 cm in addition to venous diameter was predictive for the AVF function [31, 39].

Apart from venous diameter, some authors have found an association between the presence and size of venous side branches and AVF non-maturation. Wong et al. [33] suggested that a side branch <5 cm away from the planned anastomosis may impair AVF function, whereas Beathard et al. [15] have stressed the importance of the size of the venous side branches. In the aforementioned studies non-maturation was more likely in the event of a large venous side branch. Turmel-Rodrigues et al. [14], in contrast, state that venous side branches are of no importance and only come into play in the presence of a venous outflow stenosis.

Dynamic parameters to characterize upper extremity veins include flow volume and velocity measurements as well as assessment of flow wave changes due to respiratory maneuvers [4]. The capacity of superficial veins to dilate due to venous congestion (also known as compliance) has been reported to be higher within a group of patients in whom AVF creation was successful compared with patients with AVF that failed to mature (diameter increase 48% versus 11.8% at a congestion pressure of 50 mmHg) [9]. Forearm superficial venous compliance measurements, however, have been reported to be poorly reproducible due to poor reproducibility of venous diameters at low venous congestion pressures [38]. The clinical value of forearm superficial venous compliance measurements is therefore considered of little if any use.

Digital subtraction angiography

Digital subtraction angiography (DSA), using X-ray techniques and iodinated contrast-media is considered the standard of reference for assessment of upper extremity arteries and veins. A drawback, however, is that acquired data are two-dimensional (2D) projections and analysis is limited by the number and quality of the acquired projections. Furthermore, the clinical applicability in the work-up of ESRD patients is limited due to the use of iodinated nephrotoxic contrast-media that may lead to temporary or permanent deterioration of residual renal function in up to 20% of the cases [40]. Further deterioration of renal function should be avoided because loss of residual renal function is associated with higher morbidity and mortality rates [41, 42]. Another important reason to avoid the use of iodinated contrast media is the chance of renal function recovery, even after initiation of hemodialysis therapy. In addition, residual kidney functions such as secretion of organic acids and various endocrine functions cannot be provided by dialysis and should therefore be preserved as long as possible [41]. An additional obvious drawback of DSA is the invasive nature of the examination, especially for the depiction of the arterial system.

Geoffroy et al. [40] reported on the safety and clinical performance of IA-DSA using gadolinium chelates in ESRD patients and concluded that they were well tolerated with minimal impairment of renal function (mean 3%, <12% in all cases). However, others reported serious complications associated with the off-label use of available gadolinium chelates at doses exceeding 0.3 mmol/kg [43–46]. The high osmolality of contrast media is a pathogenetic factor in contrast induced nephropathy (CIN) that potentially leads to further renal function impairment. CIN is more likely in patients with pre-existing impaired renal function [43, 44, 46, 47]. Gadolinium chelates are hypertonic and the osmolality of commercially available gadolinium chelates is two- to seven-times that of plasma [43, 46]. To prevent CIN due to hypertonicity the dose should not exceed 0.3 mmol/kg. Older patients and patients with lower baseline creatinine clearance, diabetic nephropathy and low haemoglobin and albumin levels are at increased risk for developing gadolinium induced CIN, with a reported incidence between 0 and 11% [40, 47].

Another alternative to iodinated contrast-media and gadolinium chelates is carbon dioxide (CO2) gas injection [40]. In a recent report Heye et. al. [48] demonstrated that CO2 venography is an acceptable alternative (sensitivity, 97%; specificity, 85%; accuracy, 95%) for assessment of upper-extremity and central veins in patients with contraindications to conventional venography with iodinated contrast material. In Fig. 4, a CO2 venogram and corresponding conventional venogram are shown. However, CO2 injections may cause pain and the CO2 contrast technique can lead to stenosis grade overestimation [40]. The use of CO2 contrast may furthermore lead to serious complications, such as brain gas embolism, pulmonary embolism or acute cardiac arrest [40].

Fig. 4
figure 4

Example of venograms obtained in a single patient by CO2 venography (a) and by conventional venography using iodinated contrast media (b). (Images were kindly provided by Dr. S. Heye, from the department of radiology, University Hospitals Gasthuisberg, Leuven, Belgium)

Full size image

Arterial assessment

Upper extremity arteriography is traditionally performed by intra-arterial injection of contrast-media. Arterial access can be achieved by femoral or brachial puncture. In patients with ESRD the brachial artery approach is generally avoided because it can be painful, it may jeopardize distal perfusion and thereby maturation and function of a vascular access created distal to the puncture site. Thrombosis of the brachial artery is a serious complication due to brachial punctures for catheter access in up to 7% [49–51]. Contralateral venous injection of contrast media is a less invasive alternative with fewer complications. However, in order to achieve sufficient arterial enhancement, this technique requires a high contrast dose which is unacceptable in ESRD patients due to the potential further deterioration of residual renal function.

Venous assessment

Central venous stenosis and obstruction occur frequently after central venous catheter insertion or placement of pacemaker wires, however, often patients remain asymptomatic. Imaging of the subclavian—innominate—and superior caval veins prior to vascular access creation is important because 40% of patients with a history of central venous catheters have central venous stenosis or obstruction [52]. Examples of central venous obstructions due to pacemaker wires and central venous catheter use are shown in Figs. 5 and 6.

Fig. 5
figure 5

Two consecutive images obtained by digital subtraction angiography of a central venous obstruction (arrow) due to pacemaker wires. Arrowheads point at collateral draining veins

Full size image
Fig. 6
figure 6

Two consecutive images obtained by digital subtraction angiography of a central venous obstruction (arrow) due to central venous catheter use. Arrowheads point at collateral draining veins

Full size image

Venography by cannulation of an ipsilateral dorsal hand vein allows imaging of the entire cephalic or basilic veins from the hand up to the confluence of the basilic and brachial veins into the subclavian vein. Although the superficial veins of the upper extremity are connected to each other at multiple levels, the puncture site will limit venous opacification to the draining vein of the puncture site only. It is important to avoid puncturing veins proximal to the distal radius in order to preserve draining veins for future access use. In our experience, the use of a proximal tourniquet inflated to 60 mmHg enables depiction of collateral veins and improves assessment of venous diameter because of dilatation. Good inter-observer correlation coefficients have been reported for assessment of venous quality prior to access creation using conventional X-ray venography techniques and gadolinium chelates as contrast media (kappa values, cephalic vein: forearm 0.65, arm 0.88; basilic vein: forearm 0.72; arm 0.64). Opacification of upper extremity veins is generally adequate although opacification of central veins can remain problematic, even after administration of increased contrast volumes [40].

Magnetic resonance angiography (MRA)

Ongoing technical improvements have made MRA an important and valuable imaging modality in recent years in the preoperative workup of patients with ESRD. With the introduction of contrast-enhanced MRA techniques, the role of non-enhanced time-of-flight (TOF) MRA has declined as it is prone to artifacts and stenoses are frequently overestimated [53]. Because of this reason, we only discuss contrast-enhanced MRA (CE-MRA).

CE-MRA protocols enable image acquisition with high spatial (sub-millimeter voxel-size) and temporal resolutions (<20 s per dynamic scan) with good to excellent image quality [54, 55]. For arterial imaging 0.2–0.3 mmol/kg 0.5 M extracellular gadolinium chelate contrast medium is injected in a contralateral antecubital or dorsal hand vein. For venous imaging diluted (1:15) contrast media is injected in an ipsilateral dorsal hand vein.

Because intra-thoracic vessels are prone to movement during respiration, patients should hold their breath for about 15–20 s depending on scan protocol and technical factors related to system performance. Contrast medium is injected at speeds up to 3.0 ml/s, followed by 25 ml of saline flush. Spatial resolution in recent reports is typically in the order of 1.0×1.0×1.2 mm3 (cranio-caudal/frequency direction×left-right/phase-encoding directions×antero-posterior/slice direction) [54]. Using this approach for imaging, upper extremity arteries and veins can be visualized with high accuracy [54]. Because the average upper extremity length of an adult is about 70–80 cm, depicting the entire upper extremity requires imaging of at least two fields-of-view because of the limited MR-bore length. An example of a preoperative CE-MR arteriogram and venogram of an ESRD-patient is shown in Fig. 7.

Fig. 7
figure 7

Example of selective CE-MRA of upper extremity arteries (a) and veins (b), by contralateral and ipsipateral injection of contrast media, respectively. Images of the entire upper extremity vasculature were acquired in four consecutive phases. First distal arteries (forearm), second proximal arteries (upper arm and thorax), third proximal veins (upper arm and thorax) and fourth distal veins (forearm). (sa subclavian artery, ba brachial artery, ra radial artery, ua ulnar artery, sv subclavian vein, cv cephalic vein, bv basilic vein)

Full size image

Synchronization of peak arterial or venous contrast concentration with sampling of central k-space profiles is very important in order to obtain selective imaging of arteries or veins, respectively [56]. This is typically done by performing a timing sequence with 1–2 ml contrast medium prior to the contrast-enhanced acquisition, or by the use of real-time bolus monitoring software. In order to improve both spatial and temporal resolution multi-element surface coils can be used to enable parallel imaging techniques [57]. Apart from their parallel imaging capabilities these coils greatly improve vessel-to-background contrast.

Arterial assessment

Planken et al. [54] reported a multi-phase approach using multiple dynamic scans that resulted in good to excellent subjective image quality images of upper extremity arteries down to the wrist. Assessment of the arterial palmar arch and digital arteries remains cumbersome, however, with the currently acquired spatial resolutions, except when using timed arterial compression [55]. An important advance in this regard is the recent clinical introduction of blood pool contrast media which enables ultra-high spatial resolution scans for assessment of the palmar arch and digital arteries (Fig. 8) [58].

Fig. 8
figure 8

Ultra high resolution (acquired voxel size: 0.4×0.4×0.4 mm3) steady state CEMRA of the arterial palmar arch and the proximal digital arteries, 45 s after injection of a blood pool agent. Although, both arteries and veins are equally opacified during steady state and separation of arteries and veins can be difficult, the arterial palmar arch (arrowheads) can be identified as well as accompanying veins (arrow)

Full size image

Although CE-MRA is believed to be highly accurate for detection of arterial stenoses and obstructions there are no studies about the accuracy and reproducibility for detection of thoracic and upper extremity arterial stenoses in ESRD patients. A typical example of a subclavian artery stenosis depicted by both CE-MRA and DSA is shown in Fig. 9.

Fig. 9
figure 9

Typical example of a subclavian artery stenosis, initially detected by CE-MRA (a) and confirmed by DSA (b). The arrow indicates the location of the stenosis

Full size image

Venous assessment

Reported CE-MR venography techniques use either direct injection of diluted contrast-media (1:15 to 1:25 ml of gadolinium chelate in saline solution) in the ipsilateral extremity or contralateral intravenous injection of non-diluted contrast-media, and acquisition during delayed venous enhancement after initial arterial first pass [59, 60]. Both techniques have their strengths and weaknesses. Direct venography with application of a proximal blood pressure cuff inflated to 60 mmHg yields better vessel opacification with lower contrast dose compared with the contralateral injection approach. Examples of a central venous stenosis and obstruction by direct MR venography are shown in Figs. 10 and 11.

Fig. 10
figure 10

Example of a direct MR venogram of a patient with a central venous stenosis. The arrowhead points at the venous outflow stenosis

Full size image
Fig. 11
figure 11

Brachio-cephalic vein occlusion in a patient who had multiple central venous catheters. The arrow points at the obstructed brachio-cephalic vein in the direct CE-MR venogram. The arrowhead indicates a draining collateral vein

Full size image

Contrast-enhanced MR venography using delayed venous enhancement techniques showed relatively poor performance for detection of central venous stenosis and occlusions (sensitivity and specificity 50% and 80%) [61, 62]. Direct CE-MR venography is easy to perform, well tolerated and highly accurate for detection of venous stenosis and obstructions in the upper extremity and central veins [59, 60, 63–65]. Furthermore, direct CE-MR venography diameter measurements are more accurate compared with duplex ultrasonography when using surgical measurements as standard of reference [54]. Because of the lower contrast dose, better vessel opacification and accuracy, direct CE-MR venography seems to be the method of first choice.

Although CE-MRA is a promising and attractive modality for imaging the upper extremity arteries and veins, there are only sparse data about the clinical value in the work-up prior to vascular access creation. Future studies are needed to determine the clinical significance of adding CE-MRA to the preoperative work up prior to vascular access creation.

Safety of gadolinium containing contrast agents in patients with ESRD

The most widely used gadolinium chelates for CE-MRA purposes are gadopentate dimeglumine (Magnevist; Schering, Berlin, Germany), gadoterate dimeglumine (Dotarem; Guerbet, Aulnay, France), gadodiamide (Omniscan; GE Health, Oslo, Norway), gadoversetamide (OptiMARK; Mallinckrodt, St Louis, Mo.) and gadoteridol (ProHance; Bracco Diagnostics, Milan, Italy). The total incidence of adverse events related to gadolinium use for CE-MRA appears to be less than 5%. The incidence of any single adverse event is approximately 1%, or lower. By far the most common events are nausea, headache, and emesis [43]. When used intravenously, no detectable nephrotoxicity has been reported and the rates of adverse events are extremely low at the doses used [66–68].

Recently, however, concerns have arisen regarding the accumulation of free gadolinium in patients with renal failure [43]. During the last decade approximately 200 cases of nephrogenic systemic fibrosis (NSF), previously known as nephrogenic fibrosing dermopathy, have been reported worldwide [69, 70]. The reported clinical sings and symptoms of NSF are subacute progressive swelling of extremities followed by more proximal involvement and severe skin induration, pain, muscle restlessness and loss of skin flexibility. NSF can lead to serious physical disability and wheelchair requirement [70]. The incidence of NSF is low and the pathophysiology is unknown [70]. To date, all gadolinium chelates and gadodiamide (Omniscan) in particular, are believed to be associated with NSF [71, 72]. Because of these reported adverse events and the suspected causative relationship, the use of gadolinium-containing contrast agents should be discouraged in ESRD patients until the causative relationship is proven untrue [71, 72].

Summary and conclusions

Although history taking and physical examination are valuable in the pre-operative work-up of ESRD-patients prior to hemodialysis access creation, according to current guidelines, additional diagnostic modalities are indicated to assess arterial and venous vessel quality, or in case of suspected central venous obstruction.

The ultimate goal of preoperative assessment is to prevent non-maturation by optimal selection of the site of anastomosis and by identification and treatment of preexisting lesions before vascular access creation. However, even with the use of available diagnostic tools, non-maturation rates remain unacceptably high and standardized protocols for preoperative DUS, DSA or CE-MRA are lacking. Reported cut-off values for studied parameters are inconsistent, which might be explained by differences between research groups in measurement protocols and differences in the parameters measured.

The phenomenon of non-maturation remains poorly understood. To date, assessment of a single morphological or functional parameter has not enabled adequate prediction of postoperative AVF function for individual patients. To further unravel causes of AVF non-maturation more research is necessary. Furthermore, determination of measurement reproducibility for each parameter and each diagnostic modality is necessary before a standardized preoperative diagnostic approach can be implemented.