During the last few decades, the endovascular repair of aortic aneurysms (EVAR) has revolutionised the treatment of thoracic and abdominal aortic aneurysms. Since 1991, when Parodi et al.1 reported the first series of successful endovascular abdominal aneurysm repair in humans, enormous developments have been accomplished in techniques, materials and equipment. More delicate techniques and sophisticated materials have made possible the treatment of most difficult cases (with short proximal neck anatomy, aneurysm sac involving the origins of major arterial branches and kinking of the aorta).
Randomised comparisons with open surgery have shown that EVAR has lower peri-procedural mortality (relative risk reduction of 3.1), fewer peri-procedural complications and sustained reduced aneurysm-related mortality at four years (4% for EVAR, 7% for open repair).2,3 However, EVAR has its limitations, foremost among which is the need for re-intervention, as complication rates can be as high as 41%.2 Late complications requiring re-intervention are much less frequent, at rates of approximately 2.1–2.8%.4 The most serious complications include endoleaks, infection, graft migration and rupture.
Due to these potential problems inherent to EVAR, lifelong surveillance is currently recommended using different imaging methods. Imaging should focus on the following parameters: measurement of the aortic sac diameter, detection and classification of endoleaks and detection of any failure of the structural integrity of the endograft.5 The ideal follow-up modality should be inexpensive, widely available, reproducible and accurate, without radiation exposure.
Complications After Aortic Stent Grafting
Rupture is the most feared complication that can be encountered after EVAR;6 although it does not occur frequently (1% per year),7 due to its high mortality rate it should always be kept in mind. Predisposing factors for rupture include endoleaks (usually type I and III), stent-graft migration, disintegration and infection. Rupture of the aneurysm can occur long after the procedure, and has been encountered even in cases with proven sac shrinkage. It is believed that an adverse event such as type III or I endoleak, stent-graft disintegration or device migration results in a sudden increase of endosac pressure, thus leading to rupture.
Since the published results of the EUROpean collaborators on Stent– graft Techniques for abdominal aortic Aneurysm Repair (EUROSTAR) study,7 which reported an annual cumulative rate of rupture of around 1% per year, with recent stent-grafts and techniques/follow-up protocols the rate has declined to around 0.5% per year.8 The rupture-associated mortality rate is high (60%), regardless of the treatment option (endovascular or open surgical repair).8
An endoleak is defined as a blood flow external to the stent-graft and inside the aneurysm sac. Endoleaks can sometimes be difficult to diagnose and treat. Five types of endoleak have been described.9 Type I endoleak is caused by the absence of a seal between the endograft and the wall of the artery; blood flow originates from a stent-graft attachment site (proximal or distal). Immediate type I endoleaks are detected immediately after deployment at digital subtraction angiography (DSA), where opacification of the aneurysm sac is depicted despite endograft placement. The most frequent causes of immediate type I endoleak include angulation of the proximal or distal neck, the presence of mural thrombus or calcifications or faulty endograft dimensions. The delayed type I endoleaks can be caused by proximal or distal landing-zone enlargement, and/or endograft body or limb migration (see Figure 1). Type II endoleaks are attributed to a branch-to-branch flow, causing retrograde flow through aortic branches (for example the inferior mesenteric and the lumbar arteries) into the aneurysm sac. Type II endoleaks are the most common endoleaks encountered. The number of patent branch vessels correlates with the risk of endoleak development10 (see Figure 2). Type III endoleaks occur when there is a structural failure of the stent-graft (stent-graft fracture, holes, junctional separations seen with modular devices)9 (see Figure 3). Type IV endoleaks are caused by stent-graft porosity, while type V endoleaks (also called endotension) are diagnosed when there is continuous sac expansion although no visible sac perfusion is depicted by the imaging studies. Type I and III endoleaks require immediate treatment, while type II are usually benign and require treatment only in cases of persistent sac growth.9
Graft infection during EVAR is considered quite rare, occurring with an incidence of around 0.4%11,12 versus 1.3% during open repair.13 Graft infection usually occurs within the first four months after graft implantation. Contamination (Staphylococcus aureus) during the procedure is considered the main cause of infection.
Patients commonly present with an aorto-enteric or aorto-bronchial fistula for abdominal and thoracic aneurysms, respectively, abdominal abscess, groin fistula and septic embolisation. In around one-third of patients, the infection initially manifests with vague symptoms (malaise, fever, weight loss). Computed tomography (CT) depicts signs of graft infection: peri-aortic and retroperitoneal inflammation of varying severity, stranding oedema of the surrounding fat tissues and fluid collections. The presence of air bubbles in the aortic sac is a strong indicator of stent-graft infection. Treatment options for endograft infection include conservative therapy (antibiotics, CT-guided drainage) and surgical removal of the prosthesis (followed by extra-anatomical bypass or in situ prosthetic reconstruction). Overall mortality is high (around 27%).12,14
The incidence rate of graft migration varies considerably among studies, with a range between 5 and 45% and mean time of presentation of 20 months after the endovascular repair.8,15–18 Migration is less frequent with the newest stent-graft technology and higher with first-generation grafts and grafts without hooks. Less frequently, migration can involve cephalic migration of the limb.
Stent-graft Kink and Access-related Complications
Severely diseased, stenosed or angulated aorta/iliac arteries and stenosed aortic bifurcation (<20mm) are predisposing factors to stent-graft kink, late graft thrombosis and occlusion (due to kinking of the graft or restricted outflow).8,19 In addition, when the access arteries (femoral and iliac arteries) are stenosed and severely diseased, the risk of dissection, pseudoaneurysm formation and even rupture is high.
Therefore, careful patient selection is necessary. For patients with severely diseased or angulated arteries, open repair should be preferred. In cases when open surgery cannot be performed, intra-operative adjuncts (iliac artery angioplasty, use of aorto-mono-iliac endograft systems with femoro–femoro bypass) can be proposed.8 When severe angulation of the graft presents after deployment, balloon inflation or placement of balloon-expandable stents can help re-model the kinked endoprosthesis.
Imaging Techniques for the Detection of Complications After Endovascular Aortic Aneurysm Repair
Despite the presence of advanced imaging modalities, many physicians still consider plain radiographs the cornerstone of aortic endograft surveillance. Device integrity, migration and conformation of the graft can be studied with multiple radiography projections. Antero-posterior and lateral projections are used to evaluate possible stent-graft migration and component separation and for the detection of kinks. Oblique radiographs are also used when searching for wire fractures.
On plain radiography films, the endograft should always be placed in the centre of the radiograph to minimise geometrical distortion,20,21 and when evaluating device migration care should be given as the position of the device is subject to parallax error. Furthermore, true migration can be difficult to diagnose in the comparative radiographs if there have been changes in vertebral height, such as crush fracture or loss of disc space. The most reliable point for comparison of stent-graft position is the arterial wall itself, as represented by areas of calcification.21
Reduction in stent overlap on comparative radiographs can be due to dislocation of the modular parts of the device and can lead to limb dislocation. Endografts with proximal fixation obtained by an uncovered barbed proximal stent extending into the suprarenal aorta can be subject to separation of the top anchor stent and the upper end of the graft material. This can be easily demonstrated on radiography.21
Compared with CT images, radiographs are not subject to metallic artefacts that deteriorate the image and make detection of strut fractures difficult.9,21 Obviously, plain radiography does not provide information in terms of the aneurysm diameter and the presence of endoleaks; therefore, it cannot be used as a stand-alone modality for EVAR follow-up.21 When performing a CT examination with contrast media, the radiographs should always precede the CT angiography (CTA) to avoid obscuration of the endograft by the extracted contrast material in the collecting systems.9
Multidetector Computed Tomography Angiography
Contrast-medium-enhanced multidetector CTA (MDCTA) is the most established follow-up modality for EVAR.
The maximum aneurysm diameter can be reliably, consistently and reproducibly measured with nearly 100% sensitivity and specificity. Multiplanar reconstructions and measurement of the aneurysm size at a level perpendicular to the centre line of the vessel helps to avoid errors caused by marked aortic tortuosity.22 Some investigators have advocated aneurysm volume measurement as a follow-up parameter instead of aneurysm maximum diameter measurements.23 An increase in the size of the aneurysm is usually associated with an endoleak.
MDCTA has a 92% sensitivity and 90% specificity for endoleak detection.24 In a few cases endoleaks may escape CTA detection and cause sac enlargement.25 Classification of endoleaks is not always accurate with MDCTA; an incorrect classification rate of 14% has been reported compared with conventional DSA, while re-classifying the type of the endoleak with DSA resulted in a change in treatment in 11% of patients.26
Endoleak detection depends on the MDCTA protocol. Different combinations of unenhanced and enhanced (arterial or delayed phase) images have been proposed: single arterial phase, bi-phasic (including arterial and delayed, or non-contrast and arterial) or tri-phasic (including non-contrast, arterial and delayed). The goal is to maintain as much sensitivity and accuracy as possible with the lowest achievable radiation exposure; however, there is no consensus so far on the ideal protocol. Generally, a single-arterial-phase protocol is less accurate than a bi-phasic,27 while the tri-phasic protocol is obviously associated with the greatest radiation burden. Therefore, bi-phasic CTA is the most widely used technique for endoleak detection, but there is no universal agreement as to whether the arterial and the delayed phase or the non-contrast scan and the arterial phase should be included to obtain accurate data. Some authors have found that low-flow type II endoleak can be more frequently missed when arterial images are used instead of delayed scanning, but the presence of low-flow type II endoleaks with no associated aneurysm sac enlargement does not seem to be associated with an overall increased risk of rupture and therefore necessitate no treatment.28–30 Streak artefacts from embolisation coils can degrade images and make detection of endoleaks difficult.
In most centres, follow-up protocols include MDCTA control at months one, three, six and 12, and yearly thereafter. Total effective dose with the above protocols is around 145–204mSv over a five-year period. For a total dose of 204mSv, the risk of cancer for a patient 70 and 50 years of age is 0.60 and 1.03, respectively (one in 170 and one in 100 patients, respectively).16 Thus, the radiation dose of MDCTA is indeed a topic of concern when long-term surveillance is necessary.
In terms of CTA’s ability to detect the structural changes of the endograft, this is achievable with the currently available MDCT, although subtle non-displaced fractures cannot be identified.31,32
Unenhanced Colour Doppler and Enhanced Ultrasound
Colour Doppler ultrasonography (CDUS) has successfully been used in population screening for abdominal aortic aneurysm, and would be ideal for EVAR follow-up (it is non-invasive, widely available and inexpensive and does not carry radiation risk or nephrotoxicity). Measurement of the aortic diameter can be reliably performed with CDUS, although it has been noted that US can result in underestimation of the maximum diameter compared with CTA.33 However, CDUS performs poorly in the detection of endoleak,34,35 and according to two systematic review studies pooled sensitivity and specificity rates were 66–69% and 91–93%, respectively36,37 (see Figure 4). A specific advantage of CDUS is the detection of the direction of flow, which is important in the classification and management of endoleaks.
Recently, many investigators have concentrated on the role of contrast-enhanced US (CEUS) in the surveillance of patients after EVAR and the detection of endoleaks. US contrast agents consist of gas bubbles that are intensely echogenic and have an excellent safety profile. Evident signs of an endoleak are presence of contrast enhancement within the aneurysm sac with or without identification of the origin or the inflow and outflow collateral vessels. Time of delay between injection and sac enhancement, as well as the morphology of the enhancement (diffuse or concentrated on a pseudocavity within the thrombosed sac), may have a potential role in the origin of the endoleak. The use of recently developed agents and tissue harmonic imaging has improved the sensitivity of CEUS. Although Napoli et al.25 and Benedick et al.38 reported excellent results on the sensitivity of CEUS with detection of endoleak even in cases where the other modalities (CECT) had failed, the results of other investigators are not so encouraging.
McWilliams et al.39 reported a 50% sensitivity and relatively high false-positive rate of CEUS using CECT as the gold standard comparison technique. AbuRahma et al.35 found the overall sensitivity of CEUS for detection of endoleaks to be 68%. The detection of type II endoleaks was significantly lower (sensitivity 50% for type II versus 88% for type I).
Recently, Chaer et al.40 published an interesting approach in the domain of EVAR surveillance and the role of ultrasonography. The authors evaluated the safety of colour duplex US scanning on a specific category of patients after EVAR, specifically those with stable or shrinking aneurysm, and concluded that US-only surveillance post-EVAR was safe in this population and could be initiated early after treatment.
Generally, US imaging has specific benefits and limitations. On the one hand, the technique is convenient and safe (no radiation exposure) and inexpensive, and US agents cause no allergy or nephrotoxicity (unlike radio-opaque contrast agents). On the other hand, it is operator- and patient-dependent (obesity and bowel gas can interfere with US scanning and patient collaboration is always required), and provides no information on endograft integrity and aneurysm morphological changes (kinking).25 Furthermore, while CE imaging can be performed for analysis of one previously defined area of the aneurysmal sac, if no evidence of the site of the endoleak is present, selection of the field to image can be problematic.25
Magnetic Resonance Angiography
MRA is used as a follow-up method after EVAR, most commonly in patients with impaired renal function or a known allergy to iodinated contrast media. It has been proved that MRI can be safely used with non-ferromagnetic stent-grafts in terms of stent deflection and heating.41 Nitinol-based stents are MR-compatible with no major artefacts that could cause image deterioration.
Most studies use dynamic gadolinium-enhanced 3D and delayed 2D gradient-echo sequences. New techniques (time-resolved sequences) have been applied with good results for better characterisation of endoleak type42 (see Figure 5). It has been suggested that MRA and MDCTA can detect endoleaks with the same sensitivity.43–45 Some authors43,46 have reported that MRA can even have higher sensitivity to detect type II endoleaks compared with mono- or bi-phasic MDCTA. Cohen et al.42 found a very high agreement level (up to 97%) between MRA and DSA in terms of endoleak classification. MRA angiography can be also safely used for the follow-up of patients after thoracic aorta stent-graft.45
Generally, MRA lacks the disadvantages of CTA, such as contrast-medium- associated nephrotoxicity, potential anaphylactic reaction and ionising radiation exposure. On the other hand, gadolinium-based contrast agents have been linked to the development of nephrogenic systemic fibrosis (NSF) or nephrogenic fibrosing dermopathy (NFD). The disease has occurred in patients with moderate to end-stage renal disease after administration of gadolinium-based contrast agents.47 Patients with pacemakers and other metallic implants are unsuitable for MR surveillance.
Digital Subtraction Angiography
DSA is considered the gold standard for the detection and classification of endoleaks.26 Due to its invasive character it is usually used for better delineation of an already proven (with MDCTA or MRA) endoleak, or in cases with sac aneurysm enlargement and no apparent endoleak on MDCTA, MRA or CEUS. The main advantage of DSA is its ability to determine blood flow direction and thus differentiate type I and III from type II endoleaks. DSA should always be performed before an endoleak is characterised as type V (endotension) and before the patient is referred for open surgical repair for endotension. Finally, DSA offers the major advantage of therapeutic treatment of the proven endoleaks.
Lifelong surveillance is mandatory after EVAR in order to detect possible complications. Current strategies and modalities for the follow-up of patients after EVAR are far from satisfactory. The medical community is still on an ongoing quest for the ideal follow-up method. MDCTA is considered the gold standard for follow-up of patients after EVAR, but radiation exposure risk is a concern and the need for alternative imaging modalities,48 low-dose acquisition CT protocols49 and properly adjusted follow-up is pressing. Some authors suggest that US imaging could be advocated in patients with stable or shrinking aneurysm surveillance.40 MRA has a similar sensitivity rate to MDCTA for the detection of endoleaks, with no radiation-related exposure risk. DSA should be used for better delineation and possible treatment of an endoleak after it has been detected.