The First 150 Endovascular AAA Repairs at a Single Institution: How Steep Is the Learning Curve?

Transcription

1 J ENDOVASC THER 69 CLINICAL INVESTIGATION The First 50 Endovascular AAA Repairs at a Single Institution: How Steep Is the Learning Curve? W. Anthony Lee, MD; Yehuda G. Wolf, MD; Bradley B. Hill, MD; Paul Cipriano, MD; Thomas J. Fogarty, MD; and Christopher K. Zarins, MD Division of Vascular Surgery, Stanford University, Stanford, California, USA Purpose: To determine whether increasing experience with endovascular abdominal aortic aneurysm (AAA) repair in a single institution will result in improved outcome. Methods: A retrospective review was undertaken of 50 consecutive cases of endovascular AAA repairs performed using the AneuRx device between October 996 and April 000 in a university-based medical center. The population was divided into early and late groups of 75 patients each. Endpoints included technical success; complications; early (30-day) morbidity, mortality and rupture; endoleak at discharge and at month; early secondary intervention; proximal neck and iliac tortuosity; extender cuff placement; femoral reconstructions beyond primary repair; total fluoroscopy time; and contrast load. Results: Baseline patient and aneurysm characteristics were similar between the groups. Technical success was 98.7%; cases were converted intraprocedurally owing to difficult iliac access (early group) and a severely angulated proximal neck (late group). There was a tendency toward more frequent use of intraoperative proximal extender cuffs in the early group (% versus 4% in the late group, p0.3). Femoral reconstructions were more frequent in the early group (36% versus 9%, p0.05). While total contrast volume was similar ( 56 versus ml, p), total fluoroscopy time was significantly reduced (p0.05) between the early and late groups. Conclusions: With attention to detail and careful patient selection, successful endovascular AAA repair can be achieved with very few conversions and low perioperative mortality even during the center s early experience. Evidence indicates, however, that a learning curve definitely exists, as shown by fewer access site problems, more accurate device deployments, and decreased fluoroscopy times as proficiency is attained. J Endovasc Ther Key words: abdominal aortic aneurysms, aneurysm morphology, AneuRx stent-graft, complications The short-term benefits of endovascular AAA repair are now well-recognized with regard to shorter procedural times, lower intraoperative blood loss, decreased intensive care unit and hospital stays, lower perioperative morbidity, and faster return to baseline function., Fouryear data from the AneuRx prospective, multicenter phase II clinical trial confirmed earlier findings of perioperative mortality comparable to open surgical repair and low rates of rupture following endovascular AAA repair. 3 Since FDA approval of commercial endografts in 999, public demand for endovascular repair and the number of physicians im- Disclosure: W. Anthony Lee, Thomas J. Fogarty, and Christopher K. Zarins hold consulting arrangements with Medtronic AVE; Drs. Fogarty and Zarins own stock in Medtronic. Address for correspondence and reprints: Christopher K. Zarins, MD, Division of Vascular Surgery, Stanford University, 300 Pasteur Drive, Suite H3600, Stanford, CA USA. Fax: , zarins@stanford.edu 00 by the INTERNATIONAL SOCIETY OF ENDOVASCULAR SPECIALISTS Available at

2 70 LEARNING CURVE IN AAA ENDOGRAFT REPAIR J ENDOVASC THER planting the devices have rapidly increased. Under strict FDA guidelines, mandatory training of physicians consists of a didactic session and live-case observations. Furthermore, for both vascular surgeons and interventionists alike, endovascular aortic procedures require a unique set of cross-disciplinary technical skills that most senior operators were not exposed to during their residency. In this study, we examined a large, single institutional series over a 4-year period to determine whether increasing experience in patient selection, device implantation, and postoperative follow-up can improve clinical outcomes, which would suggest a learning curve in the performance of this endovascular procedure. Patient Population METHODS We retrospectively reviewed the medical records of 50 consecutive patients (3 men; mean age years) who underwent endovascular AAA repair with the AneuRx stent-graft system (Medtronic AVE, Santa Rosa, CA, USA). The patients were treated between October 996 and April 000 under a protocol governing a multicenter clinical trial (phases I III) of this device. According to this FDA-approved protocol, the anatomical inclusion criteria for endovascular repair were () a proximal neck diameter ranging from 6 to 6 mm, () neck length 0 mm, (3) iliac diameter 9 to 5 mm, and (4) transverse AAA diameter 50 mm or 40 mm with a documented 5-mm increase during the prior 6 months. Aneurysms were not appropriate for endograft placement if they had a proximal neck angle 60 and severe iliac tortuosity with calcifications, which precluded safe catheter access. The patients were evaluated with a combination of thin-cut spiral computed tomography (CT) with three-dimensional (3D) reconstruction (CT angiogram) and/or conventional angiography. Diameter measurements were performed with manual calipers on hard copies of axial CT images. Angulation was semiquantitatively graded as mild (30), moderate (3 to 45), and severe (46 to 60). A single observer qualitatively graded iliac tortuosity as mild, moderate, or severe. The patients most common cardiovascular risk factors (Table ) included hypertension (59%), coronary artery disease (56%), and chronic obstructive pulmonary disease (3%). The maximum mean AAA diameter was 57.8 mm with a mean proximal neck measuring.0 mm in diameter and.5 mm long. The distribution of proximal neck angulation was 74% mild, 7% moderate, and 9% severe. Similarly, the distribution of iliac tortuosity was 84% mild, 4% moderate, and % severe. Approximately 3% of the patients had one or more concomitant iliac aneurysms. During phases I and II, patients were required to be physiologically able to tolerate either an open or an endovascular procedure; this restriction was lifted for the phase III trial. Six patients were treated during the first phase of the study, while the majority underwent repair in phase II (n 46) or III (n 6). An additional 36 implantations were performed under Institutional Review Board approved compassionate use approved by the FDA. Endovascular Repair All the procedures were performed by a vascular surgeon and interventional radiologist in an operating room; a portable C-arm fluoroscopy unit (GE-OEC Medical Systems, Inc., Salt Lake City, UT, USA) with digital cineangiography capabilities was used to monitor the procedures, which were performed via bilateral femoral artery exposures using implantation techniques described previously. Prior to discharge, each patient had a spiral CT scan and biplanar abdominal radiograms. In patients with renal insufficiency, color-flow duplex ultrasound was substituted for the initial postoperative CT scan. The follow-up protocol also included a repeat imaging study (CT or duplex) at, 3, and 6 months, with 6- month intervals thereafter. Data Collection and Analysis Intraoperative details gathered during the chart review included the type of anesthesia, stent-graft size and configuration, deploy-

3 J ENDOVASC THER LEARNING CURVE IN AAA ENDOGRAFT REPAIR 7 TABLE Patient Demographics, Risk Factors, and Aneurysm Morphology Entire Cohort (n 50) Early Patients (n 75) Late Patients (n 75) p Men 3 (87%) 64 (8%) 67 (9%) Age, y Risk factors Hypertension Diabetes mellitus Coronary artery disease Cerebrovascular occlusive disease Hypercholesterolemia Arrhythmia Chronic obstructive pulmonary disease Peripheral vascular occlusive disease Congestive heart failure Chronic renal failure Aneurysm morphology Neck diameter, mm.0.7 Neck length, mm.5. Aneurysm size, mm Iliac aneurysm 34 (3%) Proximal neck tortuosity Mild 74% Moderate 7% Severe 9% Iliac artery tortuosity Mild 84% Moderate 4% Severe % Continuous data are given as mean SD. not significant. 40 (53%) (5%) 39 (5%) 4 (9%) 3 (4%) 0 (7%) (8%) (5%) 3 (7%) 5 (7%) 48 (64%) 9 (%) 45 (60%) 7 (3%) 0 (3%) 4 (3%) 5 (33%) 9 (%) (6%) 5 (7%) (3%) (3%) 7% 0% 0% 77% 5% 8% 87% 0% 3% 8% 7% % ment success, intraoperative complications, type of femoral repairs, total fluoroscopy time, contrast load, and postoperative complications. The type and status of perioperative endoleaks, the need for early secondary intervention, and the time and cause of death were also recorded. The entire study cohort was divided into early and late groups of 75 patients each. Primary endpoints included technical success; intraoperative complications; early (30-day) morbidity, mortality and rupture; endoleak status at discharge and at month; early secondary intervention; proximal neck and iliac tortuosity; extender cuff placement; femoral reconstructions beyond access site repair; total fluoroscopy time; and contrast load. Statistical analysis was performed using Statview for Windows 95/98 (Abacus Concepts, SAS Institute, Cary, NC, USA) statistical package. Chi-square analysis and paired Student t tests were performed where appropriate; statistical significance was assumed at p0.05. RESULTS The groups were comparable in age, sex distribution, aneurysm morphology, and most cardiovascular risk factors; hypercholesterolemia was more common in the late group (p0.05) (Table ). The majority (86%) of the endovascular repairs were performed under general anesthesia (Table ); spinal or epidural anesthesia was used occasionally but local anesthetic alone was not. In the 48 (98.7%) patients who had a successful procedure, caudal misdeployment (0 mm below the lowermost renal artery) or inadvertent intraprocedural slippage tend-

4 7 LEARNING CURVE IN AAA ENDOGRAFT REPAIR J ENDOVASC THER TABLE Intraoperative Data Entire Cohort (n 50) Early Patients (n 75) Late Patients (n 75) p General anesthesia Technical success Extender cuff Proximal Distal Intraoperative complications Femoral reconstruction Fluoroscopy, min Contrast load, ml 9 (86%) 98.7% 67 (89%) 98.7% 6 (83%) 98.7% (8%) 49 (33%) 9 (%) (30%) 3 (4%) 7 (36%) (8%) 4 (7%) (8%) 7 (36%) Continuous data are given as mean SD. not significant. 3 (8%) 4 (9%) ed to occur more commonly in the early group (9 versus 3 in the late group, p0.3), necessitating a proximal extender cuff to achieve secure fixation and/or seal a proximal endoleak (Table ). Adjunctive procedures, such as endarterectomy, patch angioplasty, or interposition bypass graft, were needed in significantly fewer patients (n4) in the late group compared to the early group (n7, p0.05). While the mean contrast volumes were similar between the groups, the mean fluoroscopy time was significantly shorter in the late group (p0.05). One intraoperative conversion occurred in each group, for an identical technical success rate of 98.7%. In the early group, iliac access for the delivery catheter could not be obtained due to small, atherosclerotic external iliac arteries (Fig. ). In the late group, severe neck angulation (90) prevented proper deployment of the stent-graft (Fig. ). Neither case required emergent proximal aortic control, and the patients remained hemodynamically stable throughout the transition. Both patients are alive and doing well. Each group had an 8% rate of intraoperative complications (Table 3). Correctable misdeployments were not included in this category, but unplanned hypogastric (n) or renal artery occlusions (n) and iliac injuries Figure The first surgical conversion case, which had extensive occlusive disease involving both external iliac arteries. These can now be treated with a common iliac artery conduit performed through a limited retroperitoneal exposure.

5 J ENDOVASC THER LEARNING CURVE IN AAA ENDOGRAFT REPAIR 73 TABLE 3 Intraoperative Complications Complications Early Group Late Group Renal artery occlusion Iliac artery dissection/injury Hypogastric artery occlusion Splanchnic infarcts Peripheral thromboembolic Distal aortic rupture Re-exploration Total 3 3 Figure The second surgical conversion was caused by a markedly angulated proximal neck 90; these cases should not be done and can be avoided by better patient selection. were the most common intraoperative events in both groups. None of the hypogastric artery occlusions were bilateral. There was accessory renal artery occlusion in the early group that led to renal infarction (30% of renal parenchyma involved), but the patient remained asymptomatic. In the late group, partial coverage of renal artery orifice was detected on the postoperative CT, but the vessel was still patent, and the patient s renal function had not changed. In patient in the early group, multisegmental splenic infarcts were diagnosed on the postoperative CT, but the patient was completely asymptomatic. Equal numbers of iliac artery injuries occurred in both groups; these were severe and required retroperitoneal iliac exposure and an iliofemoral bypass to restore antegrade flow. One distal aortic rupture in the late group arose from a very narrow terminal aorta, which required kissing-balloon dilation following deployment of the bifurcated stentgraft. Postoperative CT scans demonstrated a retroperitoneal hematoma without contrast extravasation. The patient remained completely asymptomatic and did not require any blood transfusions. Surgical exploration of the femoral arteries in the immediate postoperative period was necessary following acute thrombosis in one late-group patient. Successful thrombectomy with patch angioplasty was performed without sequelae. There were perioperative (30-day) deaths (Table 4), both from acute myocardial infarction in the early group. No early ruptures occurred in either group. The endoleak rates at discharge were 46% for the early group and 3% (p0.06) for the late group and at month 5% and 9%, respectively (p). Most of the endoleaks in both groups were of type II. Secondary procedures were required in 3 early-group patients: distal extender cuffs for inadequate distal fixation unrecognized at the initial procedure and urgent femoral thrombectomy with distal bypass for acute femoral thrombosis on postoperative day 3. In the late group, only patient with an acute distal type I endoleak was treated secondarily with a distal extender cuff. The number of early postoperative complications (Table 4) were similar; they included case of colon ischemia in the early group, access complications (femoral and/or brachial arteries complicated by pseudoaneurysms, nerve injury, wound infection, or lymph leaks); instances of cardiac-related sequelae (arrhythmias, myocardial infarctions, and congestive heart failure); 4 cases of respiratory insufficiency requiring prolonged oxygen support; case each of acute renal tubular necrosis and a urinary tract infection; patients with transient ischemic attacks; and individual cases of deep venous thrombosis, retroperitoneal bleed, and leukocytosis. Three patients were readmitted for other causes.

6 74 LEARNING CURVE IN AAA ENDOGRAFT REPAIR J ENDOVASC THER Entire Cohort (n 50) TABLE 4 Postoperative Data Early Patients (n 75) Late Patients (n 75) p Death Rupture (.3%) 0 (.6%) Number of patients with postoperative complications 33 (%) 8 (4%) 5 (0%) Types of complications* Colon ischemia Access site Cardiac Pulmonary Renal Cerebrovascular Readmission Others Endoleak at discharge 39% 46% 3% 0.06 Type I Type II Undetermined 3% 35% % Endoleak at month % 5% 9% Type I Type II Undetermined % 9% % Secondary interventions 4 (.7%) 3 (4%) (%) not significant. * Some patients had more than complication % 4% % 3% 0% % 5 8 % 9% 0% % 8% 0% DISCUSSION Endovascular AAA repair is one of the most significant technological advances to occur in vascular surgery over the last decade. Its rapid acceptance and widespread use shares many parallels with laparoscopic cholecystectomy during the early half of the 990s. Since FDA approval of stent-graft systems in September 999, there has been a growing demand by physicians to be trained in the new technology, which is driven by patients who want a minimally invasive alternative to traditional surgery for a life-threatening problem. Our study was motivated by the question of whether a learning curve exists in the use of this new technology, and if so, what are some improvements in clinical outcomes that can result from this learning curve? Our study involved consecutive patients undergoing placement of one type of stent-graft system at a single institution, which eliminated any confounding effects from multiple operators or interval learning of new skill sets if different devices had been used during the study period. The groups created from the patient population were well matched in all aspects, although the late group had relatively more comorbidities (.7/patient) than the early group (.36/patient, p) owing to a more liberal patient selection process once the phase II recruitment was ended. In phase III, patients were enrolled who were too sick for phase II but were anatomically suitable and could undergo anesthesia. Another reason for the greater number of comorbidities in the late group was an implicit bias to direct endovascular therapy to the relatively sicker or older patient after FDA approval, leaving healthier or younger patients to choose between conventional or endovascular repair. Both groups possessed similar aneurysm morphology, which can present particular challenges to the operator, so optimum pre-

7 J ENDOVASC THER LEARNING CURVE IN AAA ENDOGRAFT REPAIR 75 operative imaging is essential. While gross assessments of proximal neck angulation and iliac tortuosity can be made from axial images alone, volumetric rendering of axial CT data creates a virtual aortoiliac luminal cast that can be rotated in space and visualized from all angles to give the operator a better perspective of the aneurysm morphology than conventional angiograms. 4,5 Angulation and tortuosity by themselves are rarely the rate-limiting factors in successful endograft repair. Usually, the combination of proximal angulation with a short neck or iliac tortuosity with severe atherosclerosis leads to technical difficulties or failure. In this study, CT angiography with 3D reconstruction or conventional aortography was available in 60% of the patients. In some cases, preoperative aortography was not routinely performed, and patients who had spiral CT scans adequate for procedural planning from the referring hospital did not otherwise have indications for a preoperative angiogram (e.g., concomitant occlusive disease). Indeed, the threshold for repeating a CT scan with 3D reconstruction or aortography was fairly low, especially during our early experience; usually only unresolved anatomical questions, such as a short neck or accessory renal arteries, prompted repeat imaging. Therefore, of the 40% of patients who had only preoperative CT imaging, most were assumed to have only mild proximal neck angulation and iliac tortuosity. Both intraoperative conversions were related to the learning curve. In the early group, the endograft procedure would likely have been successful if performed today. The difficult iliac anatomy in this patient can now be overcome in nearly all cases using a retroperitoneal approach and anastomosis of a prosthetic conduit to the common iliac artery. In the second conversion, excessive aortic neck angulation exceeded the limits of flexibility of the delivery catheter and the physical ability to straighten the aortic neck using a stiff guidewire. Although this case was in the late group, it reflects a continuing learning curve in patient selection. We now know that proximal neck angulation 60 is a contraindication to an endovascular approach. As an operator progresses through his or her learning curve, new skills are acquired that ultimately improve performance. One important technique that was acquired to minimize the risk of intraprocedural device slippage was buttressing the stent-graft during the final steps of deployment (Fig. 3). Other more subtle techniques, such as proper positioning of the fluoroscope to minimize parallax errors (e.g., craniocaudal rotation of the image intensifier to account for lumbar lordosis and anterior displacement of the proximal aortic neck by the aneurysm) and accurately placing the endograft as close as possible to the renal arteries, contributed to the decreased need for proximal extender cuffs in the late group. One of the problems associated with the caudal misdeployments we encountered was inadvertent occlusion of the hypogastric artery due to the relatively fixed length of the device. Recent studies have demonstrated significant rates (40%) of long-term morbidity from hip and buttock claudication following unilateral hypogastric occlusion. 6 8 Based on these data and our own clinical experience, it has been our policy to attempt to revascularize these hypogastric arteries with a short external iliac hypogastric artery bypass when preoperative planning anticipates stent-graft exclusion of the hypogastric artery. Our early experience was notable for the femoral arterial injuries that can result from passage of a -F sheath in small, atherosclerotic vessels. Although no single factor can fully account for the significantly reduced incidence of adjunctive femoral artery reconstructions between the early and late groups, the combination of proximal exposure of the common femoral artery, better selection of cannulation site and sheath management, and more recently, selective use of a direct iliac conduit all likely contributed to this clinical improvement. Reduction in total fluoroscopy time was due to a greater level of awareness of radiation hazards and refinements in endovascular techniques. The rates of endoleak at discharge decreased in the second group, although the difference was not statistically significant. While some of this reduction may be due to better patient selection, most of the change was related to the number of type II endoleaks. We

8 76 LEARNING CURVE IN AAA ENDOGRAFT REPAIR J ENDOVASC THER Figure 3In the buttressing technique, retraction of the delivery sheath (A, thick arrow) just below the distal margin of the stent-graft, visualization of the bottom of the stent-graft instead of the nosecone (B, arrow), and concomitant application of slight cephalad support while pulling the runners (A, thin arrows) helped decrease by half the incidence of caudal misdeployment of the primary bifurcated stent-graft in the late group. cannot attribute this to improved technical skills, as we did not routinely embolize patent branch vessels preoperatively. Furthermore, the rate of type II endoleaks at the -month interval decreased in both groups by approximately 50%, which adds to the general impression that many of the branch vessel endoleaks seal early in the postoperative period. In conclusion, endovascular AAA repair can be performed with a very high rate of technical success, fewer femoral access problems, more accurate device deployment, and decreased fluoroscopy time as the operator acquires proficiency with fluoroscopic imaging, catheter and guidewire maneuvers, and device deployment. However, this requires careful attention to detail and interdisciplinary cooperation among interventional specialties. REFERENCES. Zarins CK, White RA, Schwarten D, et al. AneuRx stent graft versus open surgical repair of abdominal aortic aneurysms: multicenter prospective clinical trial. J Vasc Surg. 999;9: Zarins CK, Wolf YG, Lee WA, et al. Will endovascular repair replace open surgery for abdominal aortic aneurysm repair? Ann Surg. 000;3: Zarins CK, White RA, Moll FL, et al. The AneuRx stent graft: four-year results and worldwide experience 000. J Vasc Surg. 00; 33:S Rubin GD, Shiau MC, Schmidt AJ, et al. Computed tomographic angiography: historical perspective and new state-of-the-art using multi detector-row helical computed tomography. J Comput Assist Tomogr. 999;3Supp :S Armerding MD, Rubin GD, Beaulieu CF, et al. Aortic aneurysmal disease: assessment of stent-graft treatment-ct versus conventional angiography. Radiology. 000;5: Razavi MK, DeGroot M, Olcott C, et al. Internal iliac artery embolization in the stent-graft treatment of aortoiliac aneurysms: analysis of outcomes and complications. J Vasc Interv Radiol. 000;: Lee C, Kaufman JA, Fan CM, et al. Clinical outcome of internal iliac artery occlusions during endovascular treatment of aortoiliac aneurysmal diseases J Vasc Interv Radiol. 000;: Cynamon J, Lerer D, Veith FJ, et al. Hypogastric artery coil embolization prior to endoluminal repair of aneurysms and fistulas: buttock claudication, a recognized but possibly preventable complication. J Vasc Interv Radiol. 000;: