A 63-year-old man with a history of hypertension and tobacco use presented to the emergency department with acute abdominal pain and hypotension. An unenhanced CT of the abdomen and pelvis showed a ruptured abdominal aortic aneurysm (AAA) (Fig. 1A). He underwent endovascular aneurysm repair (EVAR) (Fig. 1B). The patient did well after surgery and was discharged from the hospital 1 week after the procedure.

Which imaging modalities and time frame constitute the best practices for imaging surveillance after EVAR?

An AAA is defined as a dilation of the abdominal aorta measuring greater than 3 cm. An AAA is a true aneurysm and involves all three layers of the aorta (intima, media, and adventitia) [1, 2]. Whereas aneurysms less than 4 cm in diameter have little chance of rupture, aneurysms that are 5–6 cm have a 3–15% chance of rupture [1, 3]. The in-hospital mortality rate of a ruptured AAA in the United States is approximately 53% [4]. In appropriate cases, treatment is indicated for aneurysms that measure more than 5.5 cm in men, more than 5.0 cm in women, and for patients who have symptomatic AAAs [1, 5, 6]. An AAA is typically fusiform in shape, but when saccular may be treated at a smaller diameter. The treatment of an AAA differs from false or pseudoaneurysms, which are often iatrogenic or secondary to trauma and are not enclosed by normal vessel wall [2]. Major risk factors for AAA development include male sex, age older than 60 years old, prior tobacco use, family history of AAA, hypercholesterolemia, and hypertension [7–9]. In the United States, AAA affects 1.4% of adults ages 50–84 years old and is the 15th leading cause of death overall [5].

In patients with acceptable anatomy, EVAR is preferred to open repair given its shorter recovery time and lower 30-day mortality rate [10]. The short-term benefits of EVAR over open AAA repair are counter-balanced with a potential for long-term complications, chief among which is rupture secondary to endoleak (Figs. 2 and 3, Table 1). Surveillance imaging is performed to identify endoleaks, increased sac volume, graft configuration, alignment, position, patency of limbs, and, if appropriate, to identify patients that would benefit from intervention to prevent rupture [11, 12]. The incidence of aneurysm rupture within 8 years after EVAR is low (5%), but when it does occur it is a devastating complication with a high mortality rate (67%) [13]. Other potential complications after EVAR include endograft kinking, iliac limb stenosis or thrombosis (Fig. 4), fenestration branch stenosis or thrombosis, graft infection, and graft migration (defined as movement of endograft more than 1 cm from its initial position). These complications can be detected during surveillance imaging.

After EVAR, dynamic forces on both the aneurysm sac and endograft occur throughout the patient's life that can lead to delayed endoleaks and device migration; lifelong imaging surveillance is necessary to recognize complications including delayed endoleak and sac enlargement [14, 15]. One of the most relevant changes after EVAR is expansion of the aneurysm neck by more than 3 mm [16]. Neck expansion is a major cause of both for delayed type Ia endoleak and device migration [17]. Both type Ia and Ib endoleaks have been reported to occur more than 10 years after EVAR [18]. Migration typically occurs more than 2 years after EVAR [17]. It is generally believed that delayed type Ia endoleaks are preceded by subtle neck expansion that can lead to graft displacement, expansion, and decreased graft-to-wall apposition [19–21]. Additionally, the common iliac arteries near the distal landing zones often expand and can pre-dispose to type Ib endoleak [22].

Prior studies have also highlighted the potential for interobserver and intraobserver variability with AAA [23]. One study found that interobserver measurements differed by greater than 0.5 cm in 17% of CT angiography (CTA) cases, and intraobserver variability can differ by the same amount in 1.4% of cases [23]. When CTA is compared with ultrasound, variability increases, with differences in measurements differing greater than 0.5 cm in 33% of cases [23]. The authors suggested that variability could be decreased by limiting the number of radiologists who measure the AAA, using magnification to define precise aneurysm sac borders, and keeping an open dialog between radiology and the referring physicians regarding the definition of AAA diameter [23]. Other studies have suggested that each AAA should be monitored using maximum anteroposterior and maximum transverse dimensions to reduce measurement variability [24].

Although EVAR had previously only been performed in favorable aneurysm necks, many physicians are using EVAR even in aneurysms with a hostile neck, though performance of EVAR in these cases may be outside the device-specific instructions for use (Table 2). To treat difficult necks, different endovascular approaches can be tried, including the snorkel or chimney technique, suprarenal fixation, and fenestrated endo-grafts [25]. One commonly observed complication after the use of the chimney technique is the development of gutter leaks, which are located between the fabric of the renal artery stents and aortic stent-graft. Although these are technically considered a type Ia endoleak, they have a more benign course than classic type Ia endoleaks. Nearly one-third of chimney EVARs develop a type Ia gutter endoleak; however, they rarely lead to sac enlargement and no intervention was needed in 96.7% of cases [25, 26] (Fig. 5). Gutter leaks can be classified into patterns A, B, or C according to procedural characteristics and imaging findings. Pattern A is a result of excessive graft oversizing by more than 30%, which can cause infolding of the graft leading to infolding of the superior portion of the endograft, causing a type Ia endoleak [27]. Pattern B is a result of insufficient graft oversizing (< 20%), causing a poor seal [27]. Pattern C is a result of inadequate seal length, causing possible migration and type Ia endoleak [27]. Overall, delayed type Ia endoleaks occur twice as frequently in aneurysms with a hostile neck configuration than in aneurysms with a favorable neck [28]. Additionally, aneurysm-related mortality is nearly nine times greater in patients who have aneurysms with hostile neck anatomy who undergo EVAR than in those with favorable neck anatomy who undergo EVAR (Fig. 6). More active surveillance protocol for these patients might be beneficial.

Whereas current guidelines recommend imaging surveillance, the evidence for performing surveillance is varied and controversial. Some studies have supported surveillance, finding that 10% of patients benefit from imaging surveillance after EVAR [29]. Others have shown no benefit to surveillance; and one study showed patients with incomplete surveillance had no statistical difference in aneurysm-related mortality (ARM) [30]. Still other studies have shown worse survival in patients who underwent surveillance [31]. This article reviews the literature and existing recommendations for imaging surveillance after EVAR and previews devices and technologies that may help.

After EVAR was first introduced in the early 1990s, as part of the pivotal trials the FDA recommended CTA at 1, 6, and 12 months and then annually thereafter to continue indefinitely [5, 40]. As endograft technology, imaging technology, and our understanding of adverse events after EVAR continue to evolve, such a surveillance program is no longer needed or recommended. Follow-up surveillance should be individualized to the patient, taking into account patient demographics, preprocedure imaging characteristics of the aneurysm, and appearance on the 1-month CTA. Today, both American and European vascular surgery societies recommend less frequent imaging than these original recommendations (Fig. 10).

Several issues warrant further research, including the value of direct sac pressure monitors for surveillance after EVAR; whether patients would benefit from surveillance after EVAR that is specific to their imaging findings (e.g., if short aneurysm necks necessitate more frequent surveillance with CTA); and whether imaging should be tailored to the specific stent-graft used, patient demographics, and findings on the first CTA after EVAR. Other issues for further investigation include whether artificial intelligence or other software could identify EVAR at risk for developing endoleak or aneurysm expansion through detection of subtle changes between examinations, thereby identifying patients who need greater surveillance; whether contrast-enhanced ultrasound could become the sole imaging modality for surveillance after EVAR; and whether concern over CIN should dictate the use of iodinated contrast material in imaging surveillance after EVAR.

After EVAR, we make the following recommendations. All patients should be evaluated with triple-phase CTA within 30 days after EVAR. All patients should undergo lifelong imaging surveillance. Patients with a shrinking aneurysm and no endoleak on initial CTA can be monitored with annual duplex ultrasound. During ultrasound surveillance, if sac enlargement or a new endoleak is identified, the patient should proceed to multiphase CTA or MRA for further evaluation.

MRA should be used instead of CTA in these scenarios: patients with iodinated contrast material allergy, patients at risk for CIN, young patients in whom avoidance of ionizing radiation is desired, type V endoleaks identified by CTA, or inconclusive imaging findings on other modalities.

Patients with type I or type III endoleaks should be referred for expeditious intervention. Patients with type II and type V endoleaks should be referred for elective intervention if sac diameter has grown by more than 1 cm over 1 year.

Surveillance of patients treated with the chimney or snorkel technique does not need to differ from surveillance of those treated with regular EVAR despite the pervasiveness of type Ia endoleaks seen with the chimney technique. More frequent surveillance in patients who have undergone EVAR who have aneurysms with a hostile neck should be considered. Dose reduction techniques such as using dual-energy scanning to create virtual unenhanced images should be done whenever available.

We thank Jennie Williams for her assistance in creating the medical illustration of Figure 3.