We retrospectively reviewed the clinical charts of 16 consecutive patients who were referred to Shanghai Delta Health Hospital for surgical intervention for post-TEVAR AEF between June 2019 and June 2024. All patients had undergone initial TEVAR at other institutions and subsequently developed AEF with stent-related infection. Due to the referral nature of the cohort, complete baseline data on the initial TEVAR were not fully available. Based on the available records, the initial indication for TEVAR was type B aortic dissection for the majority of the patients, while for a minority, it was a thoracic aortic aneurysm. All patients had proximal landing zones in zone Z2 or Z3. No patient had a history of prior aortic surgery before developing AEF. Among these patients, 15 were male and 1 was female. AEF developed after stent implantation, with a median time of 30 months (ranging from 2 weeks to 13 years). Additionally, clinical data, including the clinical symptoms, pre-existing comorbidities, preoperative antibiotic use, and classification of infectious bacteria, were all collected.

Clinical data were collected from patient records. All patients had provided written consent and agreed to have their data made available for research purposes. The local ethics committee approved this study (approval number SDH 2024 005).

AEF was diagnosed based on a comprehensive assessment of three components: (1) a clear history of fever or other infectious manifestations, such as hematemesis or dysphagia; (2) radiological evidence, including the presence of a bubble shadow around the stent on aortic CT angiography or positive findings on PET-CT when AEF could not be diagnosed; (3) confirmation of esophageal fistula location and size via preoperative esophagoscopy. A positive blood culture was considered a supportive diagnostic criterion but was not mandatory, given the high rate of prolonged preoperative antibiotic use in this referral cohort.

A two-stage surgical strategy was utilized for all patients, as illustrated in Figure 1. The patient was placed in a supine position, and a midline thoraco-abdominal incision was created, extending from the suprasternal notch to the umbilicus. For patients requiring abdominal aortic reconstruction, the incision was further extended caudally to the pubis. The abdominal aorta was exposed via a transperitoneal approach. Systemic anticoagulation was achieved with intravenous administration of heparin (100 U/kg), targeting an activated clotting time (ACT) > 400 s to ensure adequate anticoagulation during vascular anastomosis.

To evaluate the hemodynamic rationality of the surgical strategy, we performed patient-specific hemodynamic simulations before and after the procedure. The fluid mechanics processing method for this procedure comprises preoperative and postoperative processing. In preoperative processing, the patient-specific anatomical data were acquired from preoperative aortic CTA scans, followed by image preprocessing to build and initialize a vascular fluid mechanics model. Postoperative CTA scans were obtained and processed using DetecModeling software (Boea Wisdom, Zhejiang, China) to update the model, simulate postoperative hemodynamics, and evaluate the effectiveness of the intervention.

Infection status was assessed using the white blood cell count, C-reactive protein level, erythrocyte sedimentation rate, and blood culture. Antibiotic use was individualized based on preoperative and intraoperative microbiological data. For patients with preoperative positive blood cultures and confirmed sensitivity results, postoperative antibiotics were continued using the same effective regimen that had controlled fever and infectious symptoms preoperatively. For patients with negative preoperative cultures but positive intraoperative cultures, postoperative antibiotics were adjusted according to the intraoperative culture sensitivity results. For patients with no available preoperative or intraoperative culture data, empirical broad-spectrum antibiotics were administered to cover Gram-positive bacteria, Gram-negative bacteria, and fungi. All patients received a standardized antibiotic course of 6–8 weeks postoperatively.

Intravenous nutrition was initiated within 1 week after surgery, and nutritional support was then gradually transitioned to intestinal tube feeding. After confirming the healing of the esophageal fistula through follow-up esophagography, feeding was resumed, and the chest drainage tube was gradually withdrawn every 2 weeks. The patient received total aortic CTA in the outpatient department at 6 months and 1 year after hospital discharge.

During follow-up, patients were considered to have recurrent AEF if they met all three of the following criteria: (1) persistent or recurrent massive purulent drainage in the thoracic cavity; (2) positive bacterial culture of the drainage fluid with the same pathogen as identified preoperatively; and (3) confirmation of an unhealed esophageal fistula.

Severe infection was defined as septic shock requiring vasopressor support to maintain vital signs despite adequate antibiotic therapy and fluid resuscitation.

Normality of continuous variables was assessed using the Shapiro–Wilk test. Data were presented as the mean ± standard deviation (SD) for normally distributed data or the median (Q1, Q3) for non-normally distributed data. Categorical data were presented as numbers (percentages). Kaplan–Meier survival analysis was performed using the “survminer” package in R version 4.3.3. The time origin was defined as the date of the surgical procedure. The primary endpoint for overall survival was all-cause death, with endpoint dates determined as the date of death for deceased patients; surviving patients were censored at the final follow-up date. Recurrence-free survival was defined as the time from surgery to the first occurrence of AEF recurrence or all-cause death, whichever came first.

A total of 16 patients, including 15 male individuals and 1 female individual, were included in this study. The baseline characteristics of the patients are summarized in Table 1. The average age was 52.7 ± 6.7 years. Regarding the initial symptoms, fever was observed in all 16 patients (100%), and hematemesis occurred in five patients (31.25%). Three patients (18.75%) had dysphagia, a hissing sound, and black stools, respectively. Ten patients (62.5%) had a history of hypertension, three patients (18.75%) had type II diabetes, and one patient (6.25%) had an aortic root aneurysm. Bacterial culture was successful in nine patients, and seven of them were infected with Gram-negative bacteria (Table 2). AEF occurred with a median interval of 30 months (2.275, 81.75) after stent implantation. Notably, one patient underwent a Bentall operation due to the presence of an aortic root aneurysm, and this patient was the only person who required extracorporeal circulation during the surgery.

Table 3 shows that the surgery duration was 460.5 (433.5, 543.5) min, followed by a median intensive care unit (ICU) stay of 7 days (5.25, 31.75). After the surgery, the median hospital stay was 87.5 (61.5, 111) days, and the median fasting time was 75 (46, 98) days. The average duration of postoperative antibiotic administration was 7 weeks. All patients achieved effective infection control after receiving the individualized antibiotic regimen combined with surgical treatment. Notably, no patient required long-term suppressive antibiotic therapy during the follow-up period.

Postoperative complications were observed in 10 patients (62.5%, 10/16), including 1 case of cerebral ischemic stroke, 4 cases of incision infection, 1 case of artificial blood vessel rupture, 2 cases of mediastinal infection, 1 case of type A aortic dissection, and 1 case with concurrent artificial blood vessel rupture, mediastinal infection, and paraplegia. Four patients, namely, the one with the cerebral ischemic stroke, two with mediastinal infection, and the one with concurrent artificial blood vessel rupture, mediastinal infection, and paraplegia, ultimately died. Three of them succumbed to multiple organ failure, and one died from thoracic aortic rupture and hemorrhage. Notably, three out of the four fatal cases were associated with AEF recurrence (recurrence time: 2–6 weeks postoperatively). According to the Clavien–Dindo classification, the incidence of complications of grade 3 or higher was 56.3%, including four patients with grade 5, two patients with grade 4, and three patients with grade 3 complications (Table 3). The four patients who died were all of Clavien–Dindo grade 5.

Figure 5 shows the CT scans from preoperative to 5 months post-surgery of a typical patient. The preoperative CT scan revealed characteristic infection-related air bubbles. Esophagography indicated the extravasation of the contrast agent in the esophagus 2 weeks and 2 months after the operation. Esophagography performed 5 months after the operation showed no extravasation of contrast media, indicating that the esophageal fistula had healed. The median follow-up time for the remaining 12 patients was 707 (453, 1127.5) days (Table 3). The overall survival rate was 75% (95% confidence interval: 56.5%, 99.5%), and the recurrence-free survival rate was 80.8% (95% confidence interval: 63.4%, 100%) (Figure 6).

To verify the hemodynamic feasibility of the surgical strategy, patient-specific hemodynamic simulations were performed pre- and postoperatively (Figure 7). For blood flow stability (Figure 7A), the postoperative streamline map of the thoracic–abdominal artificial graft segment showed uniform blood flow distribution and no obvious swirling flow, with velocity values ranging from 0.0 to 1.5 m/s. This indicated that the extra-anatomic bypass design maintained stable laminar flow in the graft, avoiding hemodynamic disturbance. The perfusion volumes of the celiac trunk, left renal artery, and right renal artery slightly decreased postoperatively, but all values remained within the normal range for healthy individuals. The perfusion volume of the superior mesenteric artery was slightly reduced postoperatively. However, it was consistent with its preoperative level, and both values were below the healthy range. The perfusion volume of the distal abdominal aorta–iliac artery increased postoperatively, indicating improved blood supply to the lower extremities (Figure 7B). Overall, the hemodynamic analysis confirmed that the surgical approach maintained physiological organ perfusion while ensuring stable blood flow in the graft.