Abstract
Objective:
There is a growing body of literature raising concerns over the long-term durability of endovascular repair (EVAR) for abdominal aortic aneurysms (AAA), suggesting that long-term outcomes may be better after open AAA repair. However, the data investigating these long-term outcomes largely originate from early in the endovascular era and therefore do not account for increasing clinical experience and technological improvements. We investigated whether four-year outcomes after EVAR and open repair have improved over time.
Methods:
We identified all EVARs and open repairs for intact infrarenal AAA within the Vascular Quality Initiative database (2003-2018). We then stratified patients by procedure year into treatment cohorts of four years: 2003-2006, 2007-2010, 2011-2014, and 2015-2018. We used Kaplan-Meier analysis and Cox proportional hazards models to assess whether the survival following EVAR or open repair changed over time. Additionally, we propensity-matched EVAR and open repairs for each time cohort, to investigate whether the relative survival benefit of EVAR over open repair changed over time.
Results:
We included 42,293 EVARs (increasing from 549 performed between 2003-2006 to 25,433 between 2015-2018) and 5,189 open AAA repairs (increasing from 561 to 2,306). Four-year survival increased for the periods 2003-2006, 2007-2010, 2011-2014, and 2015-2018 following both EVAR (76.6% vs. 79.7% vs. 83.5% vs. 87.3%; P<.001) and open repair (82.2% vs. 85.8% vs. 87.7% vs. 88.9%; P=.026). After risk-adjustment, compared to 2003-2006, hazard of mortality up to four years after EVAR was lower in those performed between 2011-2014 (hazard ratio [HR]: 0.72; 95% confidence interval [CI]: 0.59-0.87; P=.001) and for those performed between 2015-2018 (HR: 0.56; 95%CI: 0.46-0.68; P<.001). In contrast, the risk-adjusted hazard of mortality was similar between open repair cohorts (2011-2014: HR: 0.81 [95%CI: 0.61-1.08; P=.15]; and 2015-2018: HR: 0.86 [95%CI: 0.64-1.17; P=.34]). Finally, in matched EVAR and open repairs, there was no difference in mortality in the first three cohorts, while the hazard of mortality was lower for the 2015-2018 cohort (HR: 0.65; 95%CI: 0.51-0.84; P=.001).
Conclusion:
Four-year survival following EVAR improved in patients treated in more recent years but not following open repair. This finding suggests that mid-term outcomes following EVAR are improving, perhaps due to technological improvements and increased experience, information that should be considered by surgeons and policymakers alike when evaluating the value of contemporary EVAR and open AAA repair.
Keywords: Aortic aneurysm, abdominal, endovascular procedures, mid-term, survival rate
TABLE OF CONTENT SUMMARY
In this retrospective analysis of prospectively collected data from a nationwide registry, we found that four-year survival was progressively increasing over time for EVAR but not for open repair, leading to an increasing relative benefit of EVAR compared to open repair. This finding suggests that outcomes following EVAR may be improving over time, which should be considered when evaluating the value of contemporary EVAR.
INTRODUCTION
The perioperative benefit of endovascular repair (EVAR) over open repair for infrarenal abdominal aortic aneurysms (AAA), as described in numerous clinical trials and large database studies,1-4 has led to the vast majority of AAAs currently being repaired endovascularly.5 However, the perioperative benefit following EVAR only persists until one to three years after the index procedure.2,6 This finding, in combination with the necessity of lifelong surveillance, more frequent reinterventions, and higher procedural costs, gave rise to the discussion which patients benefit from EVAR. In the UK, a draft version of the National Institute for Health and Care Excellence (NICE) guidelines, even concluded that EVAR should not be offered when open repair is suitable, mainly based on the long-term results from the EVAR-1 trial.7,8
Although EVAR has been available for over twenty years, the technology as well as the clinical experience continue to evolve. We previously demonstrated that mid-term survival improved over time following EVAR in the early endovascular era (2001-2008).6 However, it remains unclear whether this trend of improving survival following EVAR persists into the third decade of endovascular repair.
Therefore, we investigated whether four-year mortality after EVAR and open repair for infrarenal AAA has improved over the last 15 years, using a nation-wide clinical database. We hypothesized that patients undergoing EVAR presented lower mid-term mortality rates in later cohorts, resulting from the continuous research and growth in clinical experience.
METHODS
Data Source
The data in this study originated from the Society for Vascular Surgery’s Vascular Quality Initiative (VQI) clinical registry, including 525 voluntarily participating centers, divided into 18 regional quality groups. The early data in this study (between 2003-2009), originated solely from the Vascular Study Group of New England (VSGNE), as the national collaboration that resulted in the VQI was not established until 2009. This expansion from a regional to a national database has resulted in a drastic increase in participating centers and procedures captured. Participating centers collect in-hospital data of patients undergoing vascular procedures, captured in over 350 variables, including demographics, comorbidities, procedural and anatomical characteristics, as well as outcomes. Long-term mortality is captured in the VQI through its long-term follow-up module and its link to the Social Security Death Index. More information about the VQI is available at https://www.vqi.org. The institutional review board of the Beth Israel Deaconess Medical Center approved this study and waived the need for informed consent because of the de-identification of subjects within the registry.
Study Population
We identified all EVAR or open surgical repair procedures for infrarenal AAA between January 2003 and December 2018. We excluded cases for ruptures (N=4,709), cases with an unknown urgency status (N=203), and open repairs with a proximal clamp location above the infrarenal level to improve generalizability and avoid inclusion of more complex open repairs (N=4,276). Of patients entering the database multiple times, we only included the first entry and excluded later entries (N=130). One patient was excluded due to missing mortality status. Open repairs that were preceded by an EVAR within 30 days (N=36) before open repair, were included in the EVAR cohort, as those procedures may be considered as early conversions and EVAR was the initial treatment strategy. The remaining patients were stratified into even cohorts of four years based on calendar year of procedure, into those performed in the first (2003-2006), second (2007-2010), third (2011-2014), and latest time cohort (2015-2018) for both EVAR and open repair.
Clinical and Outcome Variables
Baseline characteristics, including patient demographics and comorbid conditions, as well as procedural and anatomical characteristics were compared between study groups. We calculated estimated glomerular filtration rate (eGFR; mL/min/1.72m2) according to the Chronic Kidney Disease Epidemiology Collaboration equation,9 and defined preoperative renal dysfunction as eGFR<60 or preoperative dialysis. Underweight and obesity were defined according to body mass index (BMI) as below 18.5 and above 30 kg/m2, respectively.
The primary outcome of this study was four-year survival. We compared the outcomes of this study in two ways: first, we analyzed whether there was a difference in survival between time cohorts for both EVAR and open repair separately. Secondly, we compared survival following EVAR with survival following open repair within each time cohort separately.
Statistical Analysis
Categorical variables are presented as counts and percentages and continuous variables as mean and standard deviation or median and inter quartile range (IQR), based on normality of the data. To test whether baseline, comorbid, and anatomical/procedural characteristics were increasing or decreasing over time, we used univariate linear regression modeling. This method allowed us to assess whether certain characteristics showed an increasing or decreasing linear trend per year, rather than only testing a difference between the four cohorts. We presented the outcome of this test as the slope of the regression line, hence representing the average difference (Δ) per year (in percentages for binary and in original units for continuous variables). The accompanying P-value indicated whether the regression had a non-horizontal slope, thus indicating whether the that characteristic demonstrated a trend over time.
We assessed and plotted mid-term mortality using Kaplan-Meier survival analysis up to five years after the index procedure. For all comparisons between study groups, survival was truncated at four years, corresponding to the latest time point with available survival data in each time cohort. Univariate differences in survival were tested using log-rank tests. We then used Cox proportional hazard models to identify the independent, risk-adjusted impact of treatment cohort on survival within four years. Relative differences in survival between cohorts were expressed in hazard ratios for mortality, with those below 1.0 indicating a lower hazard and thus higher survival. The large patient and outcome event numbers allowed us to adjust for a broad range of covariables that could potentially be associated with survival, including age, sex, race, Hispanic ethnicity, current or prior smoking, family history, underweight, hypertension, insulin dependent diabetes (IDDM), renal dysfunction, chronic obstructive pulmonary disease, coronary artery disease (CAD), congestive heart failure (CHF), prior aneurysm repair, symptomatic aneurysm, aneurysm diameter, iliac aneurysm, and use of statin, aspirin, beta blocker, and antiplatelet (P2Y12 antagonists). In the models analyzing EVAR, we additionally adjusted for distal device extent, and in the open repair models additionally for surgical approach, concomitant renal bypass, distal anastomosis location, hypogastric ligation, and inferior mesenteric artery management. To account for the fact that the VQI only comprised of one region (VSGNE) before 2010, we performed sensitivity analysis by only including data from that region. We also changed the reference cohort from the first to the third time cohort, thus indicating whether there were differences in mortality between the later time cohorts.
To assess differences in four-year mortality between EVAR and open repair for each time cohort, we used propensity score modeling to create matched cohorts. We first created a logistic regression model with treatment type as the dependent variable, modeling the odds for each individual to receive either EVAR or open repair. We included the same independent variables in these models as in our previously described Cox models if available for both EVAR and open. All these models had area under the operating curve (AUC) of 0.71 or higher, and Hosmer-Lemeshow goodness-of fit tests with ten groups showed that the model was correctly specified (P>.5 in all models) and propensity scores presented adequate overlap when plotted. Using the propensity scores, we matched open repair and EVAR procedures in a 1:1 greedy fashion with a caliper of 0.1. After matching, the standardized differences between all baseline and procedural variables were below the commonly used cutoff of 10%, and no differences existed in baseline characteristics between the groups (Supplementary Table I-IV).10,11 After matching, we calculated the hazard ratio using univariate Cox models.
All analyses were performed using Stata 15.1 (StataCorp, College station, Texas, USA), P-values <.05 were considered statistically significant, and tests were two-sided.
RESULTS
We identified a total of 47,482 individual patients undergoing AAA repair, 42,293 EVARs and 5,189 open repairs. The number of procedures per time cohort (2003-2006, 2007-2010, 2011-2014, and 2015-2018) increased from 549, 1,586, 14,725, to 25,433 for EVAR and 561, 510, 1,812, to 2,306 for open repair. The proportion of all AAAs repaired with EVAR increased between the time cohorts from 49%, 76%, 89%, to 92%. Data in the earliest cohort originated from nine centers (seven EVAR and eight open repair) and increased to 298 centers in the latest cohort (294 EVAR and 193 open repair). More than 75% of patients in the 2003-2006 and 2007-2010 cohort had follow-up longer than 60 months, the median follow-up was 60 months in the 2011-2014 cohort (IQR: 19-60) and 16 months in the 2015-2018 cohort (IQR: 7-31).
Baseline characteristics
Baseline characteristics of patients undergoing EVAR are presented in Table I. For EVAR, the median age decreased from 75 (IQR: 68-79) to 74 (67-80; P=.015) over the study period. The proportion of female patients did not change over the study period (19% in all cohorts). An increasing trend was observed for Hispanic ethnicity, current smoking, underweight, IDDM, COPD, and CHF. Conversely, a decreasing trend was observed the proportion of patients with white race, family history of AAA, renal dysfunction, CAD, and prior AAA repair. The use of statin increased from 60% in the early to 71% in the late cohort (P<.001), as did the use of antiplatelets (5.8%-14%; P<.001), while beta-blocker use decreased drastically (77%-53%; P<.001).
Table I -.
Baseline characteristics endovascular aneurysm repair
| Endovascular repair |
||||||
|---|---|---|---|---|---|---|
| 2003-2006 | 2007-2010 | 2011-2014 | 2015-2018 | Δ (per year) | P-value | |
| N | 549 | 1,586 | 14,725 | 25,433 | ||
| Age | 75 (68-79) | 74 (68-80) | 74 (68-80) | 74 (67-80) | −0.04 | .015 |
| Female gender | 19% | 19% | 19% | 19% | 0.0% | .65 |
| White race | 100% | 98% | 93% | 92% | −0.5% | <.001 |
| Hispanic ethnicity | 0.4% | 0.8% | 2.5% | 3.3% | 0.2% | <.001 |
| Current smoking | 31% | 28% | 32% | 32% | 0.2% | .047 |
| Family history of AAA | 13% | 12% | 9.0% | 7.7% | −0.4% | <.001 |
| Underweight | 5.3% | 1.2% | 2.4% | 2.7% | 0.0% | .135 |
| Obese | 1.2% | 3.2% | 3.1% | 3.1% | 0.0% | .31 |
| Hypertension | 80% | 85% | 84% | 83% | 0.0% | .51 |
| IDDM | 1.8% | 2.4% | 3.6% | 4.0% | 0.1% | <.001 |
| Renal dysfunction | 38% | 35% | 35% | 34% | −0.3% | .001 |
| COPD | 40% | 33% | 32% | 34% | 0.2% | .005 |
| CAD | 45% | 45% | 44% | 42% | −0.4% | <.001 |
| CHF | 13% | 12% | 12% | 13% | 0.2% | <.001 |
| Prior AAA repair | 2.6% | 3.2% | 3.4% | 2.3% | −0.2% | <.001 |
| Statin | 60% | 70% | 69% | 71% | 0.5% | <.001 |
| Aspirin | 65% | 71% | 65% | 65% | −0.2% | .020 |
| Beta-blocker | 77% | 72% | 57% | 53% | −1.7% | <.001 |
| Antiplatelet | 5.8% | 8.9% | 12% | 14% | 0.5% | <.001 |
AAA, abdominal aortic aneurysm; CAD, coronary artery disease; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; IQR, interquartile range; Δ, average difference per year
Boldface P values represent significance (P < .05), indicating that the Δ is statistically significantly different from zero, thus indicating a change over time
For open repair, age decreased on average with 0.14 years per study year, from 71 (IQR: 65-76) to 69 (63-75; P<.001). The proportion of female patients did not change over time (24%-22%; P=.22; Table II). Hispanic ethnicity, hypertension, IDDM, prior aneurysm repair, and use of statin or antiplatelet showed an increasing trend over the study period, while white race, positive family history of AAA, renal dysfunction, CAD, and use of aspirin and beta blockers showed decreasing trends.
Table II -.
Baseline characteristics open aneurysm repair
| Open repair |
||||||
|---|---|---|---|---|---|---|
| 2003-2006 | 2007-2010 | 2011-2014 | 2015-2018 | Δ (per year) | P-value | |
| N | 561 | 510 | 1,812 | 2,306 | ||
| Age | 71 (65-76) | 70 (64-76) | 69 (63-74) | 69 (63-75) | −.14 | <.001 |
| Female gender | 24% | 25% | 26% | 22% | −0.2% | .22 |
| White race | 100% | 98% | 94% | 91% | −0.8% | <.001 |
| Hispanic ethnicity | 0.2% | 0.6% | 2.6% | 3.6% | 0.3% | <.001 |
| Current smoking | 40% | 47% | 43% | 41% | −0.2% | .32 |
| Family history of AAA | 16% | 15% | 13% | 9.6% | −0.6% | <.001 |
| Underweight | 5.9% | 2.4% | 3.0% | 3.8% | −0.1% | .14 |
| Obese | 1.5% | 2.6% | 2.2% | 2.6% | 0.1% | .27 |
| Hypertension | 80% | 83% | 82% | 84% | 0.3% | .040 |
| IDDM | 0.7% | 0.6% | 3.2% | 3.0% | 0.2% | <.001 |
| Renal dysfunction | 36% | 29% | 30% | 28% | −0.5% | .002 |
| COPD | 37% | 30% | 32% | 32% | −0.2% | .29 |
| CAD | 45% | 41% | 36% | 36% | −0.7% | <.001 |
| CHF | 6.4% | 4.7% | 7.0% | 7.9% | 0.2% | .43 |
| Prior AAA repair | 1.1% | 3.3% | 7.1% | 7.8% | 0.6% | <.001 |
| Statin | 56% | 70% | 70% | 71% | 1.0% | <.001 |
| Aspirin | 68% | 75% | 67% | 60% | −0.9% | <.001 |
| Beta-blocker | 87% | 79% | 61% | 52% | −3.0% | <.001 |
| Antiplatelet | 4.6% | 5.5% | 7.4% | 9.9% | 0.5% | <.001 |
AAA, abdominal aortic aneurysm; CAD, coronary artery disease; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; IQR, interquartile range; Δ, average difference per year
Boldface P values represent significance (P < .05), indicating that the Δ is statistically significantly different from zero, thus indicating a change over time
Anatomical & procedural characteristics
For patients undergoing EVAR, the rate of symptomatic AAA increased over the study period (4.4%-9.2%; P=.040; Table III), while the mean aneurysm diameter decreased from 58 (±11) mm in the early cohort to 56 (±11) mm in the late cohort (P<.001). The rate of iliac aneurysms showed an overall increasing trend, although the rate was highest in the earliest cohort (27% to 26%; P=.001). Estimated blood loss (EBL) decreased (14 mL on average per year; P<.001), blood transfusions were given less frequently (6.0%-4.9%; P<.001), and the use of uni-iliac devices showed a decreasing trend over time (2.4%-5.2%; P<.001). Finally, procedural time decreased by 3.7 minutes on average per year (P<.001) and length of stay declined. Bilateral percutaneous femoral access was only captured after 2014 in the registry and increased from 55% in 2015 to 72% in 2018 (P<.001).
Table III -.
Anatomical & procedural characteristics for endovascular aneurysm repair
| Endovascular repair |
||||||
|---|---|---|---|---|---|---|
| 2003-2006 | 2007-2010 | 2011-2014 | 2015-2018 | Δ (per year) | P-value | |
| N | 549 | 1,586 | 14,725 | 25,433 | ||
| Symptomatic AAA | 4.4% | 8.2% | 9.2% | 9.2% | 0.1% | .040 |
| Maximum AAA diameter (mm) | 58 (±11) | 57 (±11) | 56 (±12) | 56 (±11) | −0.10 | <.001 |
| Iliac aneurysm | 27% | 21% | 25% | 26% | 0.3% | .001 |
| Estimated blood loss (mL) | 250 (125-400) | 200 (100-300) | 150 (75-250) | 100 (50-200) | −13.7 | <.001 |
| Intra-operative RBC transfusion | 6.0% | 8.1% | 6.6% | 4.9% | −0.4% | <.001 |
| Total Procedure Time (min) | 161 (124-211) | 142 (110-190) | 120 (91-169) | 111 (83-150) | −3.7 | <.001 |
| Postoperative LOS | 1 (1-2) | 2 (1-3) | 2 (1-3) | 1 (1-2) | −0.04 | .039 |
| LOS > days | 22% | 29% | 27% | 19% | −1.6% | <.001 |
| Uni-iliac device | 2.4% | 5.8% | 7.6% | 5.2% | −0.3% | <.001 |
AAA, abdominal aortic aneurysm; LOS, length of postoperative hospital stay; RBC, red blood cell; Δ, average difference per year
Boldface P values represent significance (P < .05), indicating that the Δ is statistically significantly different from zero, thus indicating a change over time
For open repair, the proportion of symptomatic patients increased slightly, from 8.4% in the early to 14% in the late cohort (P<.001; Table IV). Aneurysm diameter and the rate of iliac aneurysms remained constant over time. EBL increased by 32 mL on average per year (P<.001), and intraoperative transfusions trended towards higher rates in later cohorts (27%-29%; P=.074). Procedure time increased by 4.7 minutes on average per year (P<.001), while length of stay remained constant over time. The distal anastomosis was less frequently sutured to the aorta (53%-34%; P<.001) and the rate of bilateral hypogastric artery ligation decreased. Finally, the IMA was less often occluded and more frequently re-implanted, and rates of thrombo-embolectomy for inadequate limb perfusion after initial distal anastomosis completion went up from 3.0% to 8.1% (0.4% per year; P<.001).
Table IV -.
Anatomical & procedural characteristics for open aneurysm repair
| Open repair |
||||||
|---|---|---|---|---|---|---|
| 2003-2006 | 2007-2010 | 2011-2014 | 2015-2018 | Δ (per year) | P-value | |
| N | 561 | 510 | 1,812 | 2,306 | ||
| Symptomatic AAA | 8.4% | 10% | 14% | 14% | 0.5% | <.001 |
| Maximum AAA diameter (mm) | 61 (±13) | 59 (±14) | 59 (±15) | 60 (±15) | −.04 | .42 |
| Iliac aneurysm | 30% | 27% | 35% | 33% | 0.2% | .16 |
| Estimated blood loss (mL) | 1000 (700-1500) | 1000 (650-1600) | 1200 (700-2000) | 1200 (750-2000) | 32 | <.001 |
| Intra-operative RBC transfusion | 27% | 22% | 32% | 29% | 0.3% | .074 |
| Total Procedure Time (min) | 178 (142-228) | 191.5 (150-250) | 220 (171-290) | 231 (180-300) | 4.7 | <.001 |
| Postoperative LOS | 7 (6-9) | 6 (5-8) | 7 (5-9) | 7 (5-9) | 0.0 | .21 |
| LOS >10 days | 16% | 15% | 16% | 17% | 0.2% | .13 |
| Anterior incision | 84% | 90% | 82% | 82% | −0.4% | .006 |
| Distal anastomosis on aorta | 53% | 46% | 39% | 34% | −1.5% | <.001 |
| Unilateral hypogastric ligation | 7.1% | 7.6% | 7.6% | 7.7% | 0.0% | .65 |
| Bilateral hypogastric ligation | 5.0% | 5.3% | 3.6% | 3.1% | −0.2% | .004 |
| IMA Management | ||||||
| Occluded | 50% | 39% | 35% | 38% | −0.8% | <.001 |
| Ligated | 47% | 57% | 55% | 49% | −0.1% | .61 |
| Re-implanted | 3.2% | 4.1% | 11% | 13% | 0.9% | <.001 |
| Thrombo-embolectomy | 3.0% | 4.7% | 5.0% | 8.1% | 0.4% | <.001 |
AAA, abdominal aortic aneurysm; IMA, inferior mesenteric artery LOS, length of postoperative hospital stay; RBC, red blood cell; Δ, average difference per year
Boldface P values represent significance (P < .05), indicating that the Δ is statistically significantly different from zero, thus indicating a change over time
Changes in survival over time
In the 2003-2006 cohort, four-year survival following EVAR was 76.6%, followed by 79.7% in the 2007-2010 cohort, 83.5% in the 2011-2014 cohort, and 87.3% in the 2015-2018 cohort (P<.001; Figure 1A). A full overview of the survival estimates per year, including standard errors, is depicted in Supplementary Table V. After risk-adjustment, using the 2003-2006 cohort as reference, patients treated in the 2007-2010 cohort had similar a hazard of mortality (hazard ratio [HR]: 0.93; 95% confidence interval [CI]: 0.75-1.16; P=.52; Table V), while the later cohorts had a lower hazard of mortality (2011-2014, HR: 0.72; 95%CI: 0.59-0.87; P=.001; and 2015-2018, HR: 0.56; 95%CI: 0.46-0.68; P<.001). Changing the referent group to the 2011-2014 cohort, demonstrated that the 2007-2010 cohort had a higher hazard of mortality (HR: 1.29; 95%CI: 1.14-1.46; P<.001), while the 2015-2018 cohort demonstrated a lower hazard of mortality (HR: 0.78; 95%CI: 0.73-0.83; P<.001). Investigating the New England region demonstrated similar decreasing mortality over time (2007-2010: 1.03; 95%CI: 0.82-1.29; 2011-2014: 0.88; 95%CI: 0.71-1.09; and 2015-2018: 0.65; 95%CI: 0.51-0.84).
Fig 1.
Five-year survival per time cohort for (A) endovascular aneurysm repair (EVAR) and (B) open abdominal aortic aneurysm (AAA) repair. All standard errors were <10% at each time point.
Table V -.
Four-year survival per time cohort for endovascular and open repair
| Endovascular repair a |
Open repair b |
|||
|---|---|---|---|---|
| time cohort | HR (95% CI) |
P- value |
HR (95% CI) |
P- value |
| 2003-2006 | (reference) | (reference) | ||
| 2007-2010 | 0.93 (0.75-1.16) | .52 | 1.06 (0.76-1.47) | .74 |
| 2011-2014 | 0.72 (0.59-0.87) | .001 | 0.81 (0.61-1.08) | .15 |
| 2015-2018 | 0.56 (0.46-0.68) | <.001 | 0.86 (0.64-1.17) | .34 |
EVAR, endovascular aneurysm repair; HR, hazard ratio; CI, confidence interval all adjusted for age, sex, race, Hispanic ethnicity, current or prior smoking, family history, underweight, hypertension, diabetes, renal function, chronic obstructive pulmonary disease, coronary artery disease, congestive heart failure, prior aneurysm repair, symptomatic aneurysm, aneurysm diameter, iliac aneurysm, and use of statin, aspirin, beta blocker, and P2Y12 antagonist.
additionally adjusted for distal device extent
additionally adjusted for surgical approach, concomitant renal bypass, distal graft anastomosis location, hypogastric ligation, and inferior mesenteric artery management
For open repairs, unadjusted survival increased over time (2003-2006: 82.2%; 2007-2010: 85.8%; 2011-2014: 87.7%; 2015-2018: 88.9%; P=.026; Figure 1B). Following risk-adjustment, using the 2003-2006 cohort as referent, there were no differences in hazard of mortality for the 2007-2010 cohort (HR: 1.06; 95%CI: 0.76-1.47; P=.74; Table V), 2011-2014 cohort (HR: 0.81; 95%CI: 0.61-1.08; P=.15), and 2015-2018 cohort (HR: 0.86; 95%CI: 0.64-1.17; P=.34; Table V). Changing the reference time cohort in the original model and sensitivity analysis focusing on patients treated in the New England region yielded similar findings.
Differences between EVAR and open repair over time.
The unadjusted survival was lower following EVAR compared to open repair in the first three cohorts (76.7% vs. 82.2%, P=.030; 79.7% vs. 85.8, P=.003; 83.5% vs. 87.7%, P<.001) and similar in the latest cohort (87.3% vs. 88.9%, P=.35). Propensity modeling resulted in 345 matched EVAR and open repair pairs in 2003-2006, 461 pairs in 2007-2010, 1,650 pairs in 2011-2014, and 2,014 matched pairs in 2015-2018 (Supplementary table I-IV). In the first two cohorts (Figure 2A and 2B), hazard of mortality was similar for EVAR vs. open repair (HR: 1.19; 95%CI: 0.85-1.67; P=.30; and HR: 1.27; 95%CI: 0.92-1.76; P=.15). In the 2011-2014 cohort, EVAR had higher survival until nearly two years, after which the lines intersected, with no overall differences in hazard of mortality (HR: 1.16; 95%CI: 0.95-1.41; P=.16; Figure 2C). In the latest cohort, 2015-2018, we found that EVAR was associated with lower overall hazard of mortality compared to open repair (HR: 0.65; 95%CI: 0.51-0.84; P=.001; Figure 2D).
Fig 2.
Four-year survival for matched endovascular aneurysm repair (EVAR) and open repair procedures for 2003-2006 (A), 2007-2010 (B), 2011-2014 (C), and 2015-2018 (D). All standard errors were <10% at each time point.
DISCUSSION
In this nation-wide study including EVAR and open repair patients treated over the last 15 years, we demonstrated that the four-year mortality after EVAR was lower in patients treated in more recent years. Conversely, open repairs did not show a difference in survival over the study period. Consequently, in matched EVAR and open repair patients per time cohort, we found that there was no difference in survival in the first two cohorts. However, in the 2011-2014 cohort, we found that EVAR demonstrated an early benefit only, whereas EVAR in the latest cohort, 2015-2018, was associated with higher survival at all time points compared to open repair.
We previously investigated long-term survival after endovascular and open repair for AAA in Medicare beneficiaries (2001-2008).6 Four-year survival in the second half of that study was approximately 75%, comparable to our early cohorts. In that study, we also performed a sub-analysis comparing patients treated in the first half and second half of the cohort and found that the second half had higher rates of four-year survival. In the current study, we focused on investigating the differences in survival over time. We demonstrated that four-year survival after EVAR increased between 2003 and 2018 by nearly 11%, and that after risk-adjustment, patients treated in the last four years had nearly a 50% reduction in hazard of mortality up to four years compared to those treated in the early cohort. Although aneurysm diameter as well as patient age at the time of repair may be decreasing over the study period, these factors were accounted for in our multivariable models. While unadjusted survival also increased for open repair, this effect did not persist after risk-adjustment.
Although the current study was not able to investigate the cause of higher survival after EVAR, we hypothesized that the technological improvements of EVAR might contribute to this finding. Verzini et al. demonstrated in their 15-year single-center experience that older stent grafts were associated with worse long-term outcomes compared to newer stent grafts, which supports our hypothesis.12 In addition, in our previous Medicare investigation, we also demonstrated a decreasing trend in two-year rates of death and aneurysm related reinterventions following EVAR, but not following open repair.6 Although only available for the last cohort, there was a vast increase in rates of bilateral percutaneous access during EVAR, reflecting the adoption of technological improvements for this procedure. Our data also indicated that several procedural and perioperative measures, including operative time, blood loss, transfusion rates, and LOS decreased for EVAR, while all these outcomes, except LOS, demonstrated increasing trends for open repair. One straightforward explanation could be that the more complex procedures, which were not eligible for EVAR, were allocated to open repair. However, we did not see a difference over time in aneurysm diameter or rate of iliac aneurysms for open repair, and our multivariable models accounted for other factors that may represent increased complexity, such as distal graft anastomosis location and concomitant renal bypass. Another explanation could be that due to the shift from open to endovascular techniques, clinical experience increased for EVAR, while the scarcity of open repairs might have led to loss of experience, leading to a relative improvement of EVAR care delivery. Lastly, other factors, such as improved patient selection by better pre-operative risk assessment, enhanced follow-up by better identification of patients at higher risk for late adverse events, and even more adequate reinterventions may have played a role in the higher survival after EVAR.
The comparison of long-term survival between endovascular and open repair for AAA has been studied in several clinical trials and large retrospective analyses. These large trials all indicated that the perioperative benefit of EVAR over open repair was lost within 1-3 years after the index procedure.4,13,14 More recently, we confirmed that this perioperative survival benefit persisted up to three years after the index procedure within Medicare beneficiaries.6 However, all these studies have one thing in common: data originated from procedures performed between 1999-2008. EVAR received approval in the US in 1999, and therefore the period from those studies could be considered as the early experience of EVAR. Since we found a decrease in the mid-term mortality after EVAR and no difference after open repair, the consequence should theoretically be that a relative benefit of EVAR over open repair emerges, which was confirmed by our data. Although there is no statistical improvement in survival for open repair, the hazard ratios were improving, which might indicate a type II error, as our open repair cohort included fewer patients than EVAR. However, even if this effect were to become significant with inclusion of more patients, the effect was smaller compared to EVAR.
In the current study, we only analyzed the four-year survival. The UK EVAR trial 1, including patients with up to 15 years follow-up, revealed that there was no difference in mortality and aneurysm-related mortality between EVAR and open repair within the first eight years.7 However, for the period after eight years, EVAR was associated with higher mortality and aneurysm-related mortality compared to open repair. Similarly, our Medicare analysis indicated that there was no difference in reintervention or rupture rates within the first 3 years, after which EVAR was associated with higher rates compared to open repair and the difference was increasing over time through 8 years.6 Although trend analysis of more long-term outcomes than four years would be of great interest, this is currently not feasible as EVAR has been available in the US for only two decades. For now, our data indicate the principle that the mid-term survival is improving for EVAR.
A recent study also used the VQI database and demonstrated that although there was drastic increase in univariate survival between early and late treated EVAR patients, this effect disappeared entirely after adjustment.15 However, reconstruction of that multivariable model revealed a methodological error, which was verified by the first and senior authors of that study, who collaborated on this current analysis. We describe the comparison to the previous study in more detail in the supplementary appendix.
The current study should be interpreted in the context of its design. First, the study population of the VQI registry has not remained constant over time. Therefore, differences in survival could be attributed to differences in the study population. However, in our sensitivity analysis, only including patients treated in the same region, we found similar outcomes. Furthermore, even if there was better patient selection in later cohorts, it would not take away the fact that the mid-term outcomes after EVAR are still improving over time. Second, the VQI consists of centers that voluntarily participate in a quality improvement program, which may limit the generalizability of our findings to other centers. Third, as aforementioned, we were unable to investigate whether the rate of late ruptures and reinterventions changed over time. However, the VQI is currently working on a link with the Medicare claims data, and we plan to investigate whether these late events changed over time when the data is released. Finally, although we were able to adjust for a broad range of anatomical variables, specific anatomical characteristics such as neck length, diameter, and angle were not available and could therefore not be accounted for.
CONCLUSION
Four-year survival after EVAR is higher in patients treated in more recent years. However, this trend of increasing survival was not seen in patients undergoing open repair. Consequently, whereas there was no difference in survival after matched EVAR and open repairs in the early cohorts, in the most contemporary cohort, patients treated with EVAR showed higher survival rates compared to open repair. This finding suggests that mid-term outcomes following EVAR are improving with newer generations of endografts and increased experience, information that should be considered by surgeons and policymakers alike when evaluating the value of contemporary EVAR and open AAA repair.
Supplementary Material
ARTICLE HIGHLIGHTS:
Type of research:
Retrospective review of prospectively collected data from the Vascular Quality Initiative (VQI) clinical database.
Key Finding:
Four-year survival is higher in patients undergoing elective infrarenal EVAR more recently compared to those treated earlier, but not for open repair. Consequently, in matched EVAR and open patients, survival was similar in the earlier cohorts, while EVAR demonstrated higher survival in more recently treated patients.
Take Home Message:
Four-year survival rates after infrarenal EVAR are improving over time but not for open repair, leading to an increasing relative benefit of EVAR over open repair for infrarenal AAA. This is key information that should be taken into account by operating surgeons and policymakers when evaluating the value of contemporary EVAR and open AAA repair.
ACKNOWLEGDEMENT
This work was conducted with support from Harvard Catalyst ∣ The Harvard Clinical and Translational Science Center (National Center for Advancing Translational Sciences, National Institutes of Health Award UL 1TR002541) and financial contributions from Harvard University and its affiliated academic healthcare centers. The content is solely the responsibility of the authors and does not necessarily represent the official views of Harvard Catalyst, Harvard University and its affiliated academic healthcare centers, or the National Institutes of Health.
CONFLICT OF INTEREST OR FUNDING STATEMENT:
KD is supported by the Harvard-Longwood Research Training in Vascular Surgery NIH T32 Grant 5T32HL007734-22
HV is a consultant for Medtronic, WL Gore, Endologix, Abbott, Arsenal AAA;
MS is a consultant for Abbott Vascular, Cook Medical, Endologix, Medtronic, and Silk Road.
REFERENCES
- 1.Prinssen M, Verhoeven EL, Buth J, Cuypers PW, van Sambeek MR, Balm R, et al. A randomized trial comparing conventional and endovascular repair of abdominal aortic aneurysms. N Engl J Med. 2004;351(16):1607–1618. [DOI] [PubMed] [Google Scholar]
- 2.Schermerhorn ML, O’Malley AJ, Jhaveri A, Cotterill P, Pomposelli F, Landon BE. Endovascular vs. Open Repair of Abdominal Aortic Aneurysms in the Medicare Population. N Engl J Med. 2008. January 31;358(5):464–74. [DOI] [PubMed] [Google Scholar]
- 3.Greenhalgh R Comparison of endovascular aneurysm repair with open repair in patients with abdominal aortic aneurysm (EVAR trial 1), 30-day operative mortality results: randomised controlled trial. The Lancet. 2004. September 4;364(9437):843–8. [DOI] [PubMed] [Google Scholar]
- 4.Lederle. Outcomes Following Endovascular vs Open Repair of Abdominal Aortic Aneurysm: A Randomized Trial. Surv Anesthesiol. 2010. June;54(3):124–5. [DOI] [PubMed] [Google Scholar]
- 5.Dua A, Kuy S, Lee CJ, Upchurch GR, Desai SS. Epidemiology of aortic aneurysm repair in the United States from 2000 to 2010. J Vasc Surg. 2014. June 1;59(6):1512–7. [DOI] [PubMed] [Google Scholar]
- 6.Schermerhorn ML, Buck DB, O’Malley AJ, Curran T, McCallum JC, Darling J, et al. Long-Term Outcomes of Abdominal Aortic Aneurysm in the Medicare Population. N Engl J Med. 2015. July 23;373(4):328–38. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Patel R, Sweeting MJ, Powell JT, Greenhalgh RM. Endovascular versus open repair of abdominal aortic aneurysm in 15-years’ follow-up of the UK endovascular aneurysm repair trial 1 (EVAR trial 1): a randomised controlled trial. The Lancet. 2016. November 12;388(10058):2366–74. [DOI] [PubMed] [Google Scholar]
- 8.National Institute for Health and Care Excellence (NICE). Abdominal aortic aneurysm: diagnosis and management - Draft Guidelines [Internet]. 2018. Available from: https://www.nice.org.uk/guidance/indevelopment/gid-cgwave0769/documents [PubMed]
- 9.Levey AS, Stevens LA, Schmid CH, Zhang Y (Lucy), Castro AF, Feldman HI, et al. A New Equation to Estimate Glomerular Filtration Rate. Ann Intern Med. 2009. May 5;150(9):604–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Austin PC. An Introduction to Propensity Score Methods for Reducing the Effects of Confounding in Observational Studies. Multivar Behav Res. 2011. May;46(3):399–424. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Cole SR, Hernan MA. Constructing Inverse Probability Weights for Marginal Structural Models. Am J Epidemiol. 2008. July 15;168(6):656–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Verzini F, Isernia G, De Rango P, Simonte G, Parlani G, Loschi D, et al. Abdominal Aortic Endografting Beyond the Trials: A 15-Year Single-Center Experience Comparing Newer to Older Generation Stent-Grafts. J Endovasc Ther. 2014. June;21(3):439–47. [DOI] [PubMed] [Google Scholar]
- 13.De Bruin JL, Baas AF, Buth J, Prinssen M, Verhoeven EL. Long-Term Outcome of Open or Endovascular Repair of Abdominal Aortic Aneurysm. 2010; [DOI] [PubMed] [Google Scholar]
- 14.United Kingdom EVAR Trial Investigators, Greenhalgh RM, Brown LC, Powell JT, Thompson SG, Epstein D, et al. Endovascular versus open repair of abdominal aortic aneurysm. N Engl J Med. 2010. May 20;362(20):1863–71. [DOI] [PubMed] [Google Scholar]
- 15.Zarkowsky DS, Hicks CW, Bostock IC, Malas MB, Goodney PP. Extending endovascular aneurysm repair to more patients without better outcomes. J Vasc Surg. 2019. May;69(5):1412–1420.e1. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.