We retrospectively analyzed the records of all patients (n = 29) who underwent open surgery for TAAA at Miyazaki University Hospital between August 2011 and October 2017. The Institutional Review Board of Miyazaki University Hospital approved this study and waived the requirement for informed consent for participation on account of the retrospective nature of the study. General informed consent for undergoing the recommended interventions and investigations was obtained from each patient before surgery. In our hospital, TAAA patients were indicated for open surgery if they had non-dissecting aneurysm with a maximum diameter of >60 mm, dissecting aneurysm with a maximum diameter of >55 mm, infected aneurysm, or rapidly enlarging aneurysm.

Using a retroperitoneal approach, the entire thoracoabdominal aorta was exposed through the left fifth or sixth intercostal space. Cardiopulmonary bypass (CPB) was established through a femoral arterial cannula and a femoral venous cannula. Twenty-eight patients received partial CPB under mild hypothermia (rectal temperature, 32˚C - 34˚C), whereas the remaining patient received total CPB under deep hypothermia (nasopharyngeal temperature, 20˚C) because the distal aortic arch could not be adequately clamped. Following heparinization, the left lung was deflated as much as possible to prevent pulmonary hemorrhage due to manipulation of the lungs.

Figure 1 illustrates the novel surgical technique used in our hospital. Under

partial CPB, sequential aortic clamping was performed in the cranio-caudal direction to allow the viscera and intercostal arteries to remain perfused, maintaining the distal aortic perfusion pressure at 60 - 80 mmHg. The aorta was initially clamped just distal to the subclavian artery and at the middle of the thoracic segment (descending aorta). Then, the aorta was incised between the clamps, and patent upper intercostal arteries were oversewn. The proximal aorta was completely transected, and proximal anastomosis to a Dacron graft was created using 4-0 polypropylene sutures with Teflon strip reinforcement (Figure 1(A)).

Next, the aorta was clamped just proximal to the celiac artery, and the lower thoracic aorta was incised under intraoperative monitoring of spinal cord ischemia (somatosensory-evoked potential monitoring in patients treated before 2013; motor-evoked potential monitoring in patients treated from 2013 onwards). Balloon-tipped catheters were passed through the orifices of the intercostal and lumbar arteries (levels T8 through L1) in order to prevent SCI due to the steal phenomenon. Then, patent intercostal and lumbar artery orifices that were relatively large and had poor backflow [3], which was true especially in patients with evoked potential amplitude reduction to ≤50% of baseline on spinal monitoring [4] [5], were selectively reconstructed using an 8-mm side graft that had been previously anastomosed to the main graft at the back table. After completion of the anastomosis, the graft enabled pulsatile blood flow perfusion to the spinal cord (Figure 1(B) and Figure 1(C)).

The infra-renal portion of the abdominal aorta was clamped next. Pelvic and lower limb perfusion was retained as much as possible to maintain spinal cord perfusion. After the infra-renal portion of the abdominal aorta was opened, selective perfusion of the celiac, superior mesenteric, and renal arteries was achieved using 8-French balloon-tipped catheters via a side branch from the main CPB circuit. Upon creation of a distal anastomosis using 4-0 polypropylene sutures with Teflon stripreinforcement, perfusion of the lower body with native pulsatile flow was achieved, which increased spinal cord perfusion through the hypogastric arteries (Figure 1(C)). Each visceral and renal artery was trimmed and reconstructed in an end-to-end manner onto the side branches of the main graft.

Thereafter, patients with evoked potential amplitude reduction to ≤50% of baseline despite increasing blood pressure and/or initiation of cerebrospinal fluid drainage (CSFD) underwent additional reconstruction of one or two intercostal or lumbar arteries. Intraoperative spinal monitoring and CSFD were performed in 26 (90%) and 23 (79%) patients, respectively.

During perioperative management, the mean arterial pressure was maintained >80 mmHg by regulating vasopressor levels and the intravascular volume. A CSFD catheter was placed in the lumbar region the day before the surgery, and the cerebrospinal fluid was allowed to drain freely whenever its pressure exceeded 10 mmHg. CSFD was terminated at 48 - 72 h after surgery if the patient did not exhibit signs of SCI. Two patients with infected aneurysm underwent omentopexy following aggressive debridement and in situ graft replacement. We did not identify the Adamkiewicz artery (AKA) preoperatively.

All patients were followed up until November 2017. The mean follow-up duration was 33 ± 23 months (range, 1 - 69 months).

Patient characteristics, operative details, and postoperative outcome data were obtained from the clinical records. Among postoperative outcomes, we analyzed in-hospital mortality, early complications including neurological complications, the patency rate of reconstructed intercostal and lumbar arteries, aorta-related events, and late outcomes. Continuous variables are expressed as mean ± standard deviation. The overall survival rate and the rate of freedom from aortic-related death rate were calculated using a Kaplan-Meier method that considered the period between the date of surgery and the most recent follow-up.

Table 1 provides a summary of the preoperative characteristics of the 29 patients included in this study (age, 67 ± 12 years; age range, 35 - 84 years). Three patients (10%) had Marfan syndrome. According to the Crawford classification, two (7%) aneurysms were type I, eight (28%) were type II, 12 (41%) were type III, and seven (24%) were type IV. In terms of disease etiology, 15 (52%) patients had atherosclerotic aneurysm, 12 (41%) had dissecting aneurysm, and two (7%) had infected aneurysm. Seven (24%) patients had undergone prior abdominal aortic aneurysm (AAA) repair, including one who had also received endovascular replacement. Both hypogastric arteries were patent in 28 (97%) patients, whereas only one hypogastric artery was patent in the remaining patient (3%). No patient underwent emergency surgery for aortic rupture or acute aortic dissection.

Table 2 provides a summary of the operative data. The mean times spent for CPB, aortic cross-clamping, and surgery were 178 ± 67, 112 ± 48, and 644 ± 157 min, respectively. Nineteen patients (66%) underwent intercostal and lumbar artery reconstruction, with a total of 35 reconstructed intercostal and lumbar arteries (1.8 ± 0.5 arteries per patient).

Among the 26 patients who underwent intraoperative spinal monitoring, reduction of the evoked potential amplitude to ≤50% of the baselinevalue was detected in 13 patients (50%) (all with motor-evoked potential monitoring). In five

CRF chronic renal failure, HD hemodialysis, AAA abdominal aortic aneurysm. Data are given as mean ± standard deviation or frequency (percentage), as appropriate.

CPB cardiopulmonary bypass, ICA intercostal artery. Data are given as mean ± standard deviation or frequency (percentage), as appropriate.

patients, significant reductions in amplitude occurred during aortic clamping at T8 to L1. In all patients, the evoked potential amplitude recovered to >50% of the baseline value as soon as intercostal and lumbar artery reconstruction was completed and blood pressure was optimized. In eight patients, including two who received open distal anastomosis because aortic clamping was difficult, significant reductions in amplitude occurred at various times. In addition to optimization of blood pressure, initiation of CSFD, and prompt resumption of pelvic or lower limb perfusion, reconstruction of one or two additional intercostal or lumbar arteries was performed in five patients, which was effective in restoring the amplitude of evoked potentials to >50% of the baseline value. One patient with a history of open repair for AAA required direct suture repair for a duodenal injury.

No in-hospital deaths occurred among the 29 patients treated during the study period (Table 3).

Table 3 provides a summary of postoperative outcomes. No acute neurological complications such as paraplegia or paraparesis were noted. Delayed paraplegia occurred in one patient (3.4%). Specifically, a 66-year-old man with a Crawford type II dissecting aneurysm developed delayed paraplegia on postoperative day 2 because of hypotension and CSFD tube occlusion. The patient had undergone reconstruction of two intercostal arteries because of a decrease of ≥50% from the baseline motor-evoked potential during aortic clamping at T8 to L1. However, follow-up contrast-enhanced computed tomography (CT) revealed occlusion of the two reconstructed intercostal arteries. Nevertheless, the patient recovered spontaneously following blood pressure optimization and reinitiation of CSFD, regaining the ability to walk independently.

Infectious complications occurred in four patients (13.8%). Wound infection was noted in three patients, whereas the fourth patient developed peritonitis with pancreatitis. Chylothorax occurred in three patients (10.3%). One patient (3.4%) underwent reoperation because of postoperative bleeding. Other complications such as respiratory failure requiring prolonged mechanical ventilation or tracheotomy, brain infarction, or renal failure requiring new dialysis did not occur.

Two of the 19 patients who underwent reconstruction of the intercostal and

SCI spinal cord injury, HD hemodialysis.

lumbar arteries could not undergo contrast-enhanced CT because of postoperative renal dysfunction. Therefore, contrast-enhanced CT was used to evaluate the early patency rate in 17 patients (31 reconstructed intercostal and lumbar arteries). The overall patency rate was 35.5% (11/31).

Table 4 provides a summary of late outcomes. Secondary aortic infection occurred in three patients, each of whom had presented some early postoperative complications due to wound infection, peritonitis with pancreatitis, or duodenum injury. Specifically, a 40-year-old man with Marfan syndrome developed wound infection by Pseudomonas aeruginosa after surgery for a Crawford type II TAAA (Figure 2(A)). Open surgical debridement and lavage using saline were performed, resulting in complete skin closure and discharge on postoperative day 55. However, at 41 months after surgery, the patient presented with prosthetic vascular graft infection by P. aeruginosa (Figure 2(B)), which was treated by graft replacement with omental wrapping. A 63-year-old man developed peritonitis with pancreatitis after surgery for a Crawford type II TAAA. Necrosectomy combined with open drainage were performed, and the patient was discharged on postoperative day 249. However, at 40 months after surgery, the developed a secondary infection at a dilated segment of the abdominal aorta just below the prosthetic vascular graft. The infected dilated abdominal aorta was removed, and the patient received extra-anatomic axillary-bifemoral bypass grafting. Another patient, who had a duodenal injury associated with previous AAA repair, was discharged but developed a graft-duodenal fistula at 6 months

after surgery. Due to his poor general medical condition, the patient was treated conservatively but died of sepsis at 21 months after surgery.

Three (10.3%) patients underwent planned two-stage surgery. Specifically, one patient had a Crawford type II aneurysm and underwent graft replacement in the infra-renal abdominal aorta. Another patient had an infected aneurysm and underwent endovascular repair in the descending thoracic aorta. Finally, a patient with previous history of open repair for AAA underwent endovascular repair for a false aneurysm in the common iliac artery. Other aorta-related events such as rupture of a residual aortic aneurysm, unplanned repeat surgery for residual aortic aneurysm, and infection recurrence after surgery for infected aneurysm did not occur.

Three patients died during the late postoperative period (one aorta-related death, one cardiac death, and one cancer-related death). Figure 3 illustrates the curve describing survival free from aorta-related death. The overall survival rate was 86%, whereas the 3-year rate of freedom from aorta-related death was 95%.