Long-term outcome and reintervention after endovascular abdominal aortic aneurysm repair using the Zenith stent graft

Methods 

A total of 325 patients underwent elective endovascular AAA repair by using bifurcated Zenith stent grafts between October 1998 and December 2005 at the University of California–San Francisco Medical Center. Data on late-occurring (>30 days after stent-graft implantation) complications and interventions were collected prospectively. Subjects were excluded from the analysis if the operation was urgent or emergent or involved the implantation of uni-iliac, fenestrated, or multibranched devices. Follow-up included routine contrast-enhanced computed tomography (CT) and multiview abdominal radiographs at 1, 6, and 12 months and yearly thereafter. Exhaustive efforts to identify all causes of death included interviewing the patient’s relatives, reviewing pertinent hospital records, and obtaining all available death certificates.

Patient demographic information, including age, sex, aneurysm size, medical comorbidities, and creatinine levels, were collected prospectively. Before approval of the Zenith stent graft in 2003, all patients were enrolled in a single-center investigational device exemption protocol, for which one of the selection criteria specified high-risk status. All these high-risk patients had serious comorbid conditions. Data on medical comorbidities included the presence of diabetes mellitus, chronic obstructive pulmonary disease, and cardiac disease. Cardiac disease included any history of myocardial infarction, congestive heart failure, ischemic heart disease, or atrial fibrillation.

Failure of stent-graft function included failure to exclude the aneurysm from direct aortic flow (endoleak type I or III) and failure to convey blood to the iliac arteries (graft occlusion). Impending failure of stent-graft function included component separation (without endoleak) and kinking of the stent graft (without thrombosis). In terms of clinical outcome, the distinction is unimportant; they all resulted in reintervention. Those occurring more than 30 days after stent-graft implantation are combined here as cases of late failure, although they might equally well be termed late reintervention.

Cases of type II endoleak, and reintervention for type II endoleak, were analyzed separately in the absence of its contribution as a primary failure mode. Invasive treatment for a type II endoleak consisted of coil embolization of either the inferior mesenteric artery or the lumbar arteries. The selection criteria for endovascular AAA repair, methods of device insertion, and follow-up protocol did not change during the course of the study. However, the indications for treatment of a type II endoleak did change. Initially, the presence of any type II endoleak on the 1-month follow-up study was a sufficient indication for reintervention. Over the last 3 years of the study, type II endoleaks were treated only if the aneurysm size increased on follow-up CT scans.

All statistical analyses were performed by using Stata version 9.0 (StataCorp, College Station, Tex). Clinical features of patients with late failure were compared with those of the study cohort. Patients treated for a type II endoleak were compared with patients who had an untreated type II endoleak. Measured values are reported as percentages or means ± SD. The Student t test was used to compare the means of continuous variables, and the Pearson χ2 test and Fisher exact test were used to compare categorical variables. P ≤ .05 was considered statistically significant. Overall survival, freedom from late graft failure, and freedom from aneurysm-related mortality were calculated by using life-table analysis. The log-rank test was used to compare the Kaplan-Meier estimate of freedom from intervention for late failure.

Results 

The demographic features of the study cohort are displayed in Table I. The mean age in this cohort was 75.9 ± 7.4 years (range, 56.4-95.3 years), with 300 (92.3%) men and 25 (7.7%) women. The mean follow-up time was 836.4 ± 580.2 days (range, 30-2547 days). Cumulative follow-up rates were 37.2% (121 patients) at 3 years, 21.8% (71 patients) at 4 years, and 10.2% (33 patients) at 5 years. Overall, women were older than men at the time of operation (77.6 years compared with 75.7 years), but this was not statistically significant (P = .23). A total of 92 (28.3%) of the 325 patients were dead at the time of this study, and 25 patients (7.7%) were lost to follow-up.

Table I. Patient demographics of the cohort treated with the Zenith stent graft

	Variable	Data
	Age at operation (y)
	Mean ± SD	75.9±7.4
	Range	56.4-95.3
	Sex
	Male	300(92.3%)
	Female	25(7.7%)
	Preoperative Creatinine (mg/dL)
	Mean ± SD	1.2±0.5
	Range	0.6-7.1
	Follow-up (d)
	Mean ± SD	836.4±580.2
	Range	30-2547
	Aneurysm size (mm)
	Mean ± SD	59.5±9.4
	Range	32-100
	Medical comorbidities
	Diabetes mellitus	51(17.0%)
	Smoker	49(16.0%)
	Past smoker	210(64.6%)
	Cardiac disease	213(66.4%)
	COPD	102(31.8%)

COPD, Chronic obstructive pulmonary disease.

There were a total of 9 (2.8%) instances of late failure, comprising limb occlusion (n = 2), aneurysm rupture (n = 1), type III endoleak (n = 3), limb kink without occlusion (n = 1), occlusion of a renal artery (n = 1), and graft infection (n = 1). The average time after stent-graft placement to late failure was 516.7 days (range, 40-1454 days). Three deaths were attributed to late failure: one each of aneurysm rupture, stent-graft infection, and infection of a femoral-femoral bypass graft placed after limb occlusion.

Table II shows the clinical characteristics of patients who experienced late failure compared with the study cohort. The preoperative creatinine level, medical comorbidities, aneurysm size, and length of follow-up were all similar between the two groups. Those who experienced late failure were older at the time of AAA repair than those who did not (79.9 vs 75.8 years), but this did not reach statistical significance (P = .11). There was also a disproportionately higher number of women who required reintervention, but this also did not reach statistical significance (P = .15).

Table II. Patient characteristics of those requiring intervention for late failure compared with the study cohort

	Description	Late failure (n = 9)	Study cohort (n = 316)	P value
	Age at operation (y)			.11
	Mean ± SD	79.9±5.3	75.8±7.5
	Range	72.3-90.8	56.4-95.3
	Sex			.15
	Male	7(77.8%)	293(92.7%)
	Female	2(22.2%)	23(7.3%)
	Creatinine (mg/dL)			.64
	Mean ± SD	1.3±0.5	1.2±0.5
	Range	0.7-2.1	0.6-7.1
	Follow-up (d)			.73
	Mean ± SD	903.7±592.0	834.5±580.7
	Range	174-1682	30-2547
	Aneurysm size (mm)			.97
	Mean ± SD	59.3±6.9	59.5±9.4
	Range	43-67	32-100
	Medical comorbidities
	Diabetes mellitus	2(22.2%)	49(16.8%)	.65
	Cardiac disease	3(33.3%)	210(67.3%)	.07
	COPD	3(33.3%)	99(31.7%)	1.0

COPD, Chronic obstructive pulmonary disease.

The unadjusted Kaplan-Meier estimates for freedom from late failure are shown in Fig 1. Fig 1, a, shows the overall freedom from late failure in the entire cohort. Women seem to have increased rates of late failure over time compared with men, as demonstrated in Fig 1, b. However, this association is not statistically significant (P = .09 by log-rank test). Fig 2, a, shows the Kaplan-Meier estimate of all-cause mortality, and Fig 2, b, shows overall freedom from aneurysm-related mortality.

View Large Image Download to PowerPoint

Fig 1. Kaplan-Meier estimate of freedom from late failure in (A) the entire cohort and in (B) women compared with men (P = .09). The number of patients at each interval is shown at the bottom of the graph.

View Large Image Download to PowerPoint

Fig 2. Kaplan-Meier estimates of all-cause mortality (A) and aneurysm-related mortality (B). The number of patients at each interval is shown at the bottom of the graph.

There were 74 patients (23%) in our study cohort with a type II endoleak, of whom 19 underwent percutaneous intervention. The clinical characteristics of patients with treated and untreated type II endoleaks are shown in Table III. The 19 treated cases represent 5.8% of the entire study cohort. The overall intervention rate, including reintervention for late failure, is therefore 8.6% (28 patients).

Table III. Clinical features of patients treated for type II endoleak compared with those with untreated type II endoleak

	Description	Treated type II EL (n = 19)	Untreated type II EL (n = 55)	P value
	Age at operation (y)			.65
	Mean ± SD	77.9±6.2	77.1±7.2
	Range	67.2-90.3	58.9-95.3
	Sex			.06
	Male	14(73.7%)	50(90.9%)
	Female	5(26.3%)	5(9.1%)
	Creatinine (mg/dL)			.92
	Mean ± SD	1.2±0.4	1.3±0.4
	Range	0.9-2.4	0.6-2.3
	Follow-up (d)			.07
	Mean ± SD	1249.4±532.2	976.1±555.4
	Range	181-2156	34-2247
	Aneurysm size (mm)			.50
	Mean ± SD	61.2±12.5	59.6±6.8
	Range	39-93	45-79

EL, Endoleak.

Discussion 

All the cases of failure and impending failure of endovascular AAA repair necessitated treatment. All limb occlusions resulted in bypass or thrombolysis, followed by stent insertion. All type III endoleaks resulted in the insertion of additional stent-graft components. The same is not true of type II endoleak, which was treated less aggressively as the study progressed.

No patient in this study developed a type I endoleak or underwent reintervention as a result of migration. The absence of clinically relevant proximal attachment failure probably reflects the effects of the barbed proximal stent of the Zenith stent graft. Other stent-graft designs, which lack barb-mediated proximal attachment, are subject to high rates of migration,3 leading to secondary type I endoleak, reintervention, and aneurysm rupture.8

The type III endoleaks occurred on average 660 days after stent-graft insertion, thus indicating an unstable stent-graft structure or position. The implantation technique seems to have played a role. During the course of the study, we increased the degree of intercomponent overlap to a minimum of 1.5 stent body lengths, and none of the stent grafts implanted in the past 5 years has developed either component separation or type III endoleak.

Late graft occlusion occurred in 2 (0.6%) of 325 cases. Others studies have reported slightly higher rates,1, 2, 6, 9, 10 and the Zenith stent graft has been noted to perform worse in this regard than other contemporary stent grafts. Our experience has shown that additional stent support, in the form of a Wallstent (Boston Scientific, Natick, Mass), is effective prophylaxis against graft occlusion, but at-risk cases can be difficult to identify.11 Most of these Wallstents were inserted within a month of stent-graft implantation, if not at the initial operation. Only one patient underwent reintervention on the basis of the finding of impending graft limb occlusion (kinking) on routine follow-up imaging.

Large multicenter registries in Europe have shown high rates of late failure, due largely to a high incidence of stent fractures and graft erosion affecting one device, the Stentor (later Vanguard) device (Boston Scientific, Natick, Mass).1 No currently used stent grafts are so susceptible to structural degradation, but all are found to have fractures or small holes when explanted or studied with high-resolution imaging.12 The Zenith device is no exception,13 and we have seen stent fracture, barb separation, and top stent separation on routine plain abdominal radiographs. However, none has ever affected stent-graft function or led us to reintervene.

The unique feature of this study is the length of follow-up. The first cases in this series were performed in late 1998. The pivotal Zenith trial started in 2000, but only the 2-year data have as yet been published.7 This article is based on more than 3 years of follow-up in 121 patients (and more than 5 years in 33 patients). We believe this to be a sound basis for assessing long-term results.

The findings reported in this study led us to change our routine follow-up protocol after endovascular AAA repair. We still obtain CT scans but omit intravenous contrast enhancement if the 1-month postoperative CT scan demonstrates appropriate position and function and lack of an endoleak. All type II endoleaks in our study cohort were detected on the 1-month postoperative CT scan. We believe that the potential nephrotoxicity14 is not justified, given the infrequent diagnosis of treatable problems such as secondary endoleak (types I and III) and graft limb kinking. We still perform noncontrast CT scans and plain abdominal radiographs as routine follow-up to look for changes in aneurysm diameter or changes in stent-graft structure. The yield may be low, but neither test has any serious side effects.

Author contributions