- Email: [email protected]
A systematic review of infected descending thoracic aortic grafts and endografts
A systematic review of infected descending thoracic aortic grafts and endografts Andrea Kahlberg, MD,a Alessandro Grandi, MS,a Diletta Loschi, MD,a Frank Vermassen, MD, PhD,b Nathalie Moreels, MD,b Nabil Chakfé, MD, PhD,c Germano Melissano, MD,a and Roberto Chiesa, MD,a Milan, Italy; Ghent, Belgium; and Strasbourg, France
ABSTRACT Objective: The objective of this study was to collect and critically analyze the current evidence on the modalities and results of treatment of descending thoracic aortic surgical graft (SG) and endograft (EG) infection, which represents a rare but dramatic complication after both surgical and endovascular aortic repair. Methods: A comprehensive electronic health database search (PubMed/MEDLINE, Scopus, Google Scholar, and the Cochrane Library) identified all articles that were published up to October 2017 reporting on thoracic aortic SG or EG infection. Observational studies, multicenter reports, single-center series and case reports, case-control studies, and guidelines were considered eligible if reporting specific results of treatment of descending thoracic aortic SG or EG infection. Comparisons of patients presenting with SG or EG infection and between invasive and conservative treatment were performed. Odds ratio (OR) meta-analyses were run when comparative data were available. Results: Forty-three studies reporting on 233 patients with infected SG (49) or EG (184) were included. Four were multicenter studies including 107 patients, all with EG infection, associated with a fistula in 91% of cases, with a reported overall survival at 2 years of 16% to 39%. The remaining 39 single-center studies included 49 patients with SG infection and 77 with EG infection. Association with aortoesophageal fistula was significantly more common with EG (60% vs 31%; P ¼ .01). In addition, time interval from index procedure to infection was significantly shorter with EG (17 6 21 months vs 32 6 61 months; P ¼ .03). Meta-analysis showed a trend of increased 1-year mortality in patients with SG infection compared with EG infection (pooled OR, 3.6; 95% confidence interval, 0.9-14.7; P ¼ .073). Surgical management with infected graft explantation was associated with a trend toward lower 1-year mortality compared with graft preservation (pooled OR, 0.3; 95% confidence interval, 0.1-1.0; P ¼ .056). Conclusions: Thoracic aortic EG infection is likely to occur more frequently in association with aortoesophageal fistulas and in a shorter time compared with SG infection. Survival is poor in both groups, especially in patients with SG infection. Surgical treatment with graft explantation seems to be the preferable choice in fit patients. (J Vasc Surg 2018;-:1-11.) Keywords: Infection; Blood vessel prosthesis; Thoracic aorta; Surgery; Endovascular procedures; Fistula; Esophageal fistula; Respiratory fistula
Vascular prosthetic infection is a serious and often fatal event; moreover, infection of an aortic surgical graft (SG) or endograft (EG) previously implanted at the descending thoracic aortic level is considered one of the most feared complications of modern vascular surgery.1 In the last decades, the number of thoracic aortic proceduresdespecially endovasculardis continuously increasing, and infectious prosthetic complications
From the Department of Vascular Surgery, Vita-Salute University School of Medicine, San Raffaele Scientific Institute, Milana; the Department of Vascular and Thoracic Surgery, Ghent University Hospital, Ghentb; and the Department
have been observed more and more frequently, affecting 1% to 6 % of patients and entailing up to 75% mortality rates even when recognized and treated.1 Nevertheless, the evidence is poor and consensus documents are lacking in regard to the incidence, the clinical features, the therapeutic options, and above all the clinical outcome of this disease. The aim of this study was to collect and critically analyze the published evidence on the management and outcome of treatment of prosthetic infection after previous repair of the descending thoracic aorta to compare mortality of patients presenting with SG or EG infection and of patients managed conservatively or surgically.
of Vascular Surgery and Kidney Transplantation, University Hospital of Strasbourg, Strasbourg.c Author conflict of interest: none. Additional material for this article may be found online at www.jvascsurg.org. Correspondence: Andrea Kahlberg, MD, Department of Vascular Surgery, San Raffaele Scientific Institute, Vita-Salute University School of Medicine, Via Olgettina 60, 20132 Milan, Italy (e-mail: [email protected]). The editors and reviewers of this article have no relevant financial relationships to disclose per the JVS policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest. 0741-5214 Copyright Ó 2018 by the Society for Vascular Surgery. Published by Elsevier Inc. https://doi.org/10.1016/j.jvs.2018.10.108
METHODS Study design and search strategies. The review conformed to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement standards.2 An extensive electronic health database search was undertaken to identify all the articles that were published before November 2017 reporting on SG and EG infections after surgical repair or endovascular repair of thoracic or thoracoabdominal aortic disease. 1
2
Kahlberg et al
Journal of Vascular Surgery ---
The following electronic bibliographic sources were searched: PubMed/MEDLINE, Scopus, Google Scholar, and the Cochrane Library. Search strategies included MESH terms and key words: (“aorta, thoracic”[MeSH Terms] OR (“aorta”[All Fields] AND “thoracic”[All Fields]) OR “thoracic aorta”[All Fields]) AND (“infection”[MeSH Terms] OR “infection”[All Fields]) OR (“thoracic”[All Fields] AND “Stent-Graft”[All Fields]) AND (“infection”[MeSH Terms] OR “infection”[All Fields]) OR (“thoracic”[All Fields] AND “aorta”[All Fields]) OR “graft”[All Fields] OR “Stent-Graft”[All Fields] AND “infection”[All Fields] OR (“graft”[All Fields] AND “infection”[All Fields]) OR “aortoesophageal”[All Fields] AND (“fistula”[MeSH Terms] OR “fistula”[All Fields]) OR “aorto-oesophageal”[All Fields] AND (“fistula”[MeSH Terms] OR “fistula”[All Fields]) OR “aortobronchial”[All Fields] AND (“fistula”[MeSH Terms] OR “fistula”[All Fields]). In addition, the reference lists of all retrieved articles were examined for further relevant studies. The last search was run in October 2017. Articles in languages other than English were excluded. Eligibility criteria and study selection. Initial study selection and systematic review inclusion and exclusion were performed by two authors (A.G., D.L.) on the basis of the following eligibility criteria. Observational studies, multicenter reports, single-center series and case reports, case-control studies, and guidelines were included in the review analysis. No randomized clinical trial was found covering this topic. Articles reporting on descending thoracic EG infection in patients with initial associated supra-aortic trunk debranching (ie, proximal landing zone 1 or zone 2) were included. Articles reporting only on primary aortic infections (eg, mycotic aneurysm, infectious aortitis, primary aortoesophageal fistula [AEF] or aortobronchial fistula [ABF]), even if affected patients were treated with surgical or endovascular repair, were excluded. Papers limited to graft infection of the ascending aorta, of the arch, or of the infrarenal aorta and papers expressing general principles of vascular graft infection did not meet our eligibility criteria. Articles failing to provide mortality data specifically related to patients with descending thoracic or thoracoabdominal aortic graft infection (eg, reporting only about diagnostic modalities or mixing results of abdominal and thoracic infections) were excluded. When multiple publications on the same patient sample were identified or study populations overlapped, only the latest report was included unless the reported outcomes were mutually exclusive. Data extraction and study quality assessment. Each study (full-text article) was reviewed by two independent reviewers (A.K., A.G.), and the following data were extracted and inserted in a precompiled electronic database sheet: number of patients, baseline demographics,
2018
type of index procedure (surgical or endovascular), indication for index procedure, time interval to infection diagnosis, clinical presentation, type of management, perioperative mortality, and mortality at follow-up. If necessary, the article’s corresponding author was contacted to confirm or to clarify specific information. The quality of evidence was assessed at the individual study level by two authors (A.K., D.L.; Supplementary Table, online only). Summary process and results reporting. Because of the possible overlap of populations of patients with single-center reports, analysis of the results of multicenter studies was performed separately, reporting results (including follow-up survival) as described in the original articles and avoiding summary measures and comparative analysis. For single-center reports, results were reported summarizing (mean and proportions) available data extracted by each study, according to the type of infected graft (SG vs EG). Pooled mortality rate was reported at 30 days, 1 year, and 5 years. Meta-analysis was performed including only observational studies reporting results of direct comparison between given subgroups of patients (see Statistical Analysis). For all meta-analyses, the primary end point was considered the overall mortality rate at 1 year. Statistical analysis. Standard descriptive statistics were reported as numbers and percentages (categorical variables) or mean and range (continuous variables). Variables’ distribution was tested for normality by using the D’Agostino-Pearson test. Comparisons between patients presenting with SG infection or with EG infection were performed using two-tailed t-test for normally distributed continuous variables, Mann-Whitney test for non-normally distributed continuous variables, and Fisher exact test for categorical variables. Three odds ratio (OR) meta-analyses of all eligible series in accordance with the recommendations of the Metaanalysis of Observational Studies in Epidemiology group3 were performed: SG infection vs EG infection; conservative medical treatment vs surgical (any kind) treatment; and infected graft explantation vs preservation (considering only surgically treated patients). Pooled ORs were initially calculated using a fixed-effect model and reported as forest plots. Heterogeneity of included studies was assessed using the I2 statistics.4 When heterogeneity was considered high, ORs calculated under the random-effects model were preferred. Standard descriptive and comparison statistics were run using Prism 7 statistical software (GraphPad Software, Inc, La Jolla, Calif). The meta-analyses were performed with MedCalc statistical software, version 17.9.7 (MedCalc Software, Ostend, Belgium).
Journal of Vascular Surgery Volume
-,
Number
Kahlberg et al
3
-
Fig 1. Flow chart of studies identified and finally included in the review analysis.
RESULTS The initial literature search identified 1657 records, of which 239 were duplicates, and another 1200 were considered irrelevant after exclusion criteria application at the title or abstract level. Among the remaining 218 articles assessed for eligibility, 175 were excluded because of their study design (including 4 systematic reviews/metaanalyses summarizing the results of papers already individually evaluated and 10 papers reporting only results of diagnostic imaging) or because they reported insufficient data on study outcomes. Finally, 43 studies (the first one published in June 1999 and the last one in October 2017) reporting on 233 patients with infected SG (12 studies, 49 patients) or infected EG (31 studies, 184 patients) were included in the qualitative and quantitative analysis (Fig 1). Multicenter reports. Among included articles, four were based on large, multicenter retrospective series5-8 including 107 patients. All of these reported only data of patients with thoracic EG infection. No articles
providing multicenter reports on thoracic SG infection were found in the literature. Analysis of results of multicenter studies is summarized in Table I. Three studies reported the incidence and outcome of AEF and ABF after thoracic endovascular aortic repair (TEVAR)5,7,8; the fourth study reported the incidence and outcome of EG infection after both endovascular aneurysm repair (EVAR) and TEVAR.6 The mean age of considered patients was 70 years, and the male-female ratio was 3:1. The indication for primary TEVAR, when reported, was an atherosclerotic aneurysm in 69% of cases. The type of infected EG was reported in 60 of 107 cases, of which 42 had polyethylene terephthalate (PET)-based fabric and 18 had expanded polytetrafluoroethylene-based fabric. Specific graft models were Valiant in 2 cases (Medtronic, Santa Rosa, Calif), Talent in 11 cases (Medtronic), TAG in 18 cases (W. L. Gore & Associates, Flagstaff, Ariz), Zenith TX1 in 2 cases (Cook Medical, Bloomington, Ind), Zenith TX2 in 8 cases (Cook Medical), Endofit in 3 cases (Endomed, Phoenix, Ariz), Relay in 5 cases (Bolton
4
Journal of Vascular Surgery
Kahlberg et al
---
2018
Table I. Demographics, indication for index procedure, timing of infection diagnosis, clinical presentation, management strategies, and outcome of 107 patients described in multicenter case series of infections after thoracic endovascular aortic repair (TEVAR) Chiesa et al,5 2010
Smeds et al,6 2016
Czerny et al,7 2014
Czerny et al,8 2015
Total
19
26
36
26
107
Age, years
73.8 6 7.1
68
69 (56-75)
70 (60-77)
Male
16 (84)
No. of patients
20 (77)
27 (75)
17 (65)
80 (75)
28 (77)
Indication for index procedure Atherosclerotic aneurysm
13 (68)
e
Dissection, PAU, IMH
2 (11)
e
Traumatic aortic injury
1 (5)
e
e
e
Mycotic, AEF, ABF Secondary pseudoaneurysm Other
3 (16) e
Timing from index procedure to diagnosis of infection, days
327 (e)
15 (58)
56 (69)
e
6 (23)
8 (18)
e
4 (15)
5 (11)
5 (14)
e
5 (9)
e
e
1 (4)
4 (9)
e
3 (8)
e
3 (8)
540 (5-2100)
90 (30-150)
310 (28-1065)
317 (5-2100)
Clinical presentation Pain
1 (5)
17 (66)
Fever or chills
11 (58)
17 (66)
Hematemesis or hemoptysis
13 (68) 5 (26)
Shock Dyspnea
2 (10)
AEF
4 (15)
22 (31)
29 (81)
7 (27)
64 (60)
e
19 (53)
24 (92)
56 (69)
e
8 (22)
6 (23)
19 (23)
e
2 (5)
e
13 (68)
12 (46)
ABF
5 (26)
0 (0)
AEF þ ABF
5 (26)
0 (0)
Conservative
8 (42)
5 (19)
Fistula repair only
6 (32)
e
Surgical þ fistula repair
1 (5)
e
Surgical fistula repair
1 (5)
7 (27)
11 (14)
0 (0)
61 (57)
0 (0)
26 (100)
31 (29)
0 (0)
0 (0)
5 (5)
10 (28)
5 (19)
28 (26)
13 (36)
4 (15)
23 (28)
13 (36)
8 (31)
22 (27)
e
e
22 (49)
36 (100)
a
Management
21 (81)
Endo þ fistula repair
1 (5)
e
e
2 (8)
3 (7)
Endo fistula repair
2 (11)
e
e
7 (27)
9 (20)
28% (1-year)
39% (2-year)
Outcome (survival) Overall
16% (2-year)
29% (5-year)
Conservative
0% (30-day)
20% (1-year)
0% (1-year)
e
Stent, esophageal only
e
e
17% (1-year)
e
Open, esophageal only
e
e
43% (1-year)
e
Open þ esophageal
e
e
46% (1-year)
e
27% (2-year)
29% (5-year)
e
e
Open
ABF, Aortobronchial fistula; AEF, aortoesophageal fistula; IMH, intramural hematoma; PAU, penetrating aortic ulcer. Categorical variables are presented as number (%). Continuous variables are presented as mean 6 standard deviation or median (interquartile range). a Conservative, medical treatment with or without minimally invasive additional procedures; Fistula repair only, esophageal or respiratory fistula repair, without associated aortic repair; Surgical þ fistula repair, surgical open aortic repair associated with additional maneuvers to repair the fistula; Surgical fistula repair, surgical open aortic repair without additional maneuvers to repair the fistula; Endo þ fistula repair, endovascular aortic repair (TEVAR) associated with additional maneuvers to repair the fistula; Endo fistula repair, endovascular aortic repair (TEVAR) without additional maneuvers to repair the fistula.
Medical, Sunrise, Fla), and Powerlink in 1 case (Endologix, Irvine, Calif). In regard to possible etiologic factors of subsequent EG infection, only Smeds et al6 (grouping together abdominal and thoracic EVARs) mentioned an infectious complication during or after the index operation occurring in 34% of cases (urinary tract infection in 8%, groin infection in 7%,
other unspecified infection in 19%). The same group reported the occurrence of non-aorta-related secondary procedures between the index operation and the diagnosis of infection in 69 patients (34%). The mean time interval from index procedure to diagnosis of infection was 295 days (approximately 10 months). In 91% of reported patients, a fistula was
Journal of Vascular Surgery Volume
-,
Number
Kahlberg et al
5
-
present at diagnosis. Imaging modalities used to make or to confirm the diagnosis of aortic graft infection were computed tomography scan in 80 cases and magnetic resonance imaging in 1 case. In case of associated fistula, the diagnosis was made by means of esophagogastroduodenoscopy in 45 cases, aortography in 2 cases, and esophageal transit study in 3 cases. Only the study by Smeds et al6 specifically reported the results of bacteriologic findings. Perioperative blood cultures were positive in 63% of patients; intraoperative cultures identified gram-positive organisms in 22%, gram-negative organisms in 13%, and fungus in 5%. Intraoperative cultures were polymicrobial in 35% (and this was also considered a significant risk factor for morbidity and mortality in patients managed surgically) and negative in 30% of cases. A number of different strategies to treat EG infection were reported (Table I). Smeds et al6 reported the type of graft used for aortic reconstruction, as follows: a prosthetic graft in 16 cases (83% antibiotic soaked), a cryopreserved allograft in 3 cases, and a femoropopliteal neoaortoiliac system in 2 cases. Reported postoperative complications after surgical management of thoracic aortic graft infection included respiratory failure in 15 patients, bleeding in 13, multiorgan failure in 8, renal dysfunction in 5, mediastinitis in 4, and acute myocardial infarction in 3. At a mean follow-up of 25 months (range, 1-60 months), overall survival was 28% (30/107). Open aortic replacement showed a survival of 27% at 2 years in one study and of 29% at 5 years in another study. Surgical esophageal repair showed a survival of 43% at 1 year when performed alone and a survival of 46% at 1 year when associated with open aortic repair. Conservative strategies were generally associated with worse outcomes, showing a survival of 0% at 30 days in one study, 0% at 1 year in another study, and 20% at 1 year in the last study. Similarly, esophageal stenting alone showed a survival of 17% at 1 year. Single-center reports. The remaining 39 articles reported single-center experience: 11 were case series with >5 patients,9-19 3 were case series with #5 patients,20-22 and 25 were case reports,23-47 with a total of 126 patients included in this analysis. Among these, 49 patients presented with thoracic aortic SG infection (41 descending thoracic grafts and 8 thoracoabdominal grafts), and 77 presented with thoracic aortic EG infection. Among EG infection cases, the type of infected EG was reported in 28 of 77 cases, of which 20 had PET-based fabric and 8 had expanded polytetrafluoroethylene-based fabric. Specific graft models were Valiant in 9 cases (Medtronic), Talent in 7 cases (Medtronic), TAG in 7 cases (W. L. Gore), Zenith TX2 in 2 cases (Cook Medical), Seal thoracic stent graft in 1 case (S&G Biotech Inc, Seoul, Korea), GianturcoRosch Z stent in 1 case (assembled at authors’ institution),
and Bard custom-made thoracic stent graft in 1 case (C.R. Bard, Tempe, AZ). Demographics, data on clinical presentation, and outcomes of the two groups are presented in Table II. Eight papers mentioned some infectious complications occurring during the perioperative period of the index procedure. In particular, 7 papers9,27,28,37,39,40,45 (including a total population of 17 patients) reported pulmonary infection occurring in 9 patients, septicemia in 2, urinary tract infection in 1, coagulopathy in 1, and early duodenalparaprosthetic fistula in 1 open thoracoabdominal repair case. Another paper,14 reporting on 22 patients with EG infection (but mixing abdominal and thoracic EVARs), described 11 patients (50%) experiencing perioperative septic complications. Secondary vascular procedures occurring between the index procedure and the diagnosis of infection were reported in five cases, including one TEVAR after open thoracoabdominal aortic repair due to proximal disease progression, two cases of embolization of a type I endoleak, one bronchial stenting due to aneurysm sac compression after TEVAR, and one emergent visceral surgical debranching for thoracoabdominal disease progression. Nonvascular procedures were reported in five patients, including two cases of pulmonary resection associated with aortic replacement, one case of belowknee amputation during the interval, one cholecystectomy, and one case of multiple hospitalizations requiring placement of central venous catheters. In SG, the mean time interval from index procedure to diagnosis of infection was 32.2 months, significantly longer compared with EG (17.1 months; P ¼ .03). Association with AEF was significantly less common with SG than with EG (31% vs 60%, respectively; P ¼ .01). Symptoms at presentation, pain and fever or chills, were significantly more common with SG (73% vs 35% [P < .001] and 84% vs 61% [P ¼ .02], respectively), whereas hematemesis was significantly more common with EG (42% vs 19%; P ¼ .02). Imaging modalities used to make or to confirm the original diagnosis of aortic graft infection were computed tomography scan in 51 cases, fluorodeoxyglucose positron emission tomography in 3 cases, and magnetic resonance imaging in 1 case. In case of associated fistula, the diagnosis was made by means of esophagogastroduodenoscopy in 29 cases, aortography in 5 cases, and esophageal transit study in 4 cases. Specific antimicrobial findings were reported in 18 papers (53 isolated microorganisms in 49 patients). In almost all cases, findings were based on blood cultures. Most commonly isolated micro-organisms were Staphylococcus aureus (19 patients, methicillin resistant [MRSA] in 7) and Streptococcus spp (12 patients); other isolated microorganisms were Escherichia coli (5), Pseudomonas aeruginosa (4), Candida albicans (4), Enterococcus spp (3), Klebsiella spp (2), Lactobacillus spp (1), Haemophilus influenzae (1), Serratia spp (1), and
6
Journal of Vascular Surgery
Kahlberg et al
---
2018
Table II. Comparison of demographics, timing of infection diagnosis, clinical presentation, management, and mortality between patients with previous thoracic open aortic repair and patients with previous thoracic endovascular aortic repair (TEVAR), among the 126 patients described in single-center case series or case reports SG infection, previous open repair (n ¼ 49)
EG infection, previous TEVAR, (n ¼ 77)
Availablea
Positiveb
%c
Male sex
30
25
Age, years, mean 6 SD
47
62.5 6 11.4
76
64.9 6 12.2
.17
Timing from index procedure, months, mean 6 SD
31
32.2 6 61.1
52
17.1 6 21.1
.03
83
Availablea
Positiveb
41
30
%c 73
P .39
Clinical presentation AEF
32
10
31
60
36
60
.01
ABF
32
4
13
60
11
18
.56
AEF þ ABF
32
0
0
60
1
2
Pain
37
27
73
48
17
35
<.001
Fever or chills
37
31
84
61
37
61
.02
Hematemesis
37
7
19
52
22
42
Hemoptysis
37
6
16
52
8
15
1.0
.02 1.0
Shock
37
4
11
52
8
15
Dyspnea
24
0
0
46
1
2
1.0
.75
Sepsis
24
2
8
55
6
11
1.0
Conservative
49
1
2
77
14
18.2
.02
Fistula repair only
49
1
2
77
10
12.9
.06
Surgical þ fistula repair
49
e
e
77
2
2.6
.52
Surgical fistula repair
49
2
4.5
77
7
9.1
.48
Endo þ fistula repair
49
12
27
77
23
29.9
.69
Endo fistula repair
49
25
52
77
16
20.8
.02
TEVAR / open
49
6
12.5
77
5
6.5
.35
30-day
48
16
33
73
19
26
1-year
47
27
57
67
36
54
.71
5-year
34
28
82
52
42
81
1.0
Managementd
Mortality .42
ABF, Aortobronchial fistula; AEF, aortoesophageal fistula; EG, endograft; SD, standard deviation; SG, surgical graft. Numbers in bold are considered statistically significant. a The number of patients for whom the specific information was available in the included studies. b The number of affected patients for each single item. c Percentages of positive patients of the available ones. d Conservative, medical treatment with or without minimally invasive additional procedures; Fistula repair only, esophageal or respiratory fistula repair, without associated aortic repair; Surgical þ fistula repair, surgical open aortic repair associated with additional maneuvers to repair the fistula; Surgical fistula repair, surgical open aortic repair without additional maneuvers to repair the fistula; Endo þ fistula repair, endovascular aortic repair (TEVAR) associated with additional maneuvers to repair the fistula; Endo fistula repair, endovascular aortic repair (TEVAR) without additional maneuvers to repair the fistula; TEVAR / open, endovascular aortic repair to obtain hemodynamic stability followed by open surgical aortic repair.
Staphylococcus epidermidis (1). Thirty-day mortality in patients with documented antimicrobial growth at culture was 14% (7/49), 5 of which were positive for MRSA. For MRSA-positive patients, 30-day mortality was 71%; for methicillin-sensitive S. aureus-positive patients, 9%; and for Streptococcus-positive patients, 8%. No significant difference in outcome was noted between patients with isolation of a single agent and patients with polymicrobial isolation. Conservative treatment, mainly represented by antibiotic therapy, sometimes associated with percutaneous
drainage of fluid collections or flushing, was performed in 2% of patients with SG (1/49) and 17% with EG (13/77). This approach resulted in a mortality rate of 100% at 30 days in the SG infection group and 38%, 75%, and 100% in the EG infection group at 30 days, 1 year, and 5 years, respectively. In the remaining patients, different surgical strategies were used, including aggressive open aortic repair with or without associated fistula repair, TEVAR with or without associated fistula repair, and esophageal or bronchial-pulmonary repair alone. Seventy-four patients
Journal of Vascular Surgery Volume
-,
Number
Kahlberg et al
7
-
Fig 2. Pooled odds ratio (OR) meta-analysis comparing 1-year mortality among patients with surgical graft (SG) infection and patients with endograft (EG) infection. CI, Confidence interval.
received in situ aortic repair, using a standard PET graft in 40 cases (antibiotic soaked in 15 cases), a cryopreserved allograft in 22 cases, a xenopericardial tube graft in 9 cases, a silver-coated PET graft in 2 cases, and an autologous vein graft in 1 case. Reported postoperative complications included respiratory failure in 22 patients, persistent sepsis in 13, multiorgan failure in 13, bleeding in 13, cardiac complications in 7, renal dysfunction in 5, paraplegia in 5, mediastinitis in 5, vocal cord paralysis in 3, esophageal leak in 2, wound infection in 2, recurrent aortopulmonary fistula in 1, encephalopathy in 1, and cholecystitis in 1. Global 30-day, 1-year, and 5-year mortality was 33%, 57%, and 82% in patients with SG infection and 26%, 54%, and 81% in patients with EG infection, respectively.
DISCUSSION Published scientific evidence on thoracic aortic prosthetic infection is definitely scarce. This is mainly due to the following reasons: d
d
d
d
Meta-analyses results. The OR meta-analysis of SG infection vs EG infection included five comparative studies (Fig 2). The statistical heterogeneity was low (I2 ¼ 2%; P ¼ .38). Even if it was not statistically significant, a slight tendency to increased overall mortality rate at 1 year was noted in patients with SG infection (pooled OR, 3.6; 95% confidence interval [CI], 0.9-14.7; P ¼ .073). The OR meta-analysis of conservative treatment vs surgical treatment included five comparative studies (Fig 3). The statistical heterogeneity was low (I2 ¼ 5%; P ¼ .37). No significant difference in overall mortality rate at 1 year was noted between conservative and surgical treatment (pooled OR, 0.6; 95% CI, 0.1-2.3; P ¼ .425). Finally, the OR meta-analysis of infected graft explantation vs preservation included seven comparative studies (Fig 4). The statistical heterogeneity was low (I2 ¼ 0%; P ¼ .42). A slight trend toward lower mortality rate at 1-year outcome was noted in patients with infected graft explantation (pooled OR, 0.3; 95% CI, 0.1-1.0; P ¼ .056).
Relative rarity of this complication. The majority of published papers are case reports or small series. This is probably also a cause of underestimation because of the lack of single-center reports by small centers experiencing few cases and not reporting them. Frequent late occurrence. Several patients might receive the diagnosis and treatment of infection years later in a different center from the index procedure. Heterogeneity in clinical presentation. Reports are often mixing elective and emergent cases, with or without associated AEF or ABF fistula. Heterogeneity in treatment strategies. Published papers usually include a number of different pharmacologic, surgical, and endovascular therapeutic options. Surgical strategies and timing modalities are mostly chosen according to the surgeon’s experience and the patient’s condition.
Infection of a thoracic aortic prosthesis is, however, a morbid condition, and mortality often exceeds 50% and can be as high as 100% for untreated cases.5-8 There are no prospective randomized trials concerning the treatment options for this condition, and the presence of many different variations makes it even more difficult to draw sound conclusions that may apply to the individual case. In this paper, we considered for analysis the original studies that referred to cases with an SG or EG involvement and not the reviews as a whole. However, a brief discussion of these reviews is provided here. In the last two decades, TEVAR has been gradually replacing open intervention with surgical replacement
8
Kahlberg et al
Journal of Vascular Surgery ---
2018
Fig 3. Pooled odds ratio (OR) meta-analysis comparing 1-year mortality among patients treated conservatively and patients treated surgically (any kind of surgical treatment). CI, Confidence interval.
Fig 4. Pooled odds ratio (OR) meta-analysis comparing 1-year mortality among patients with infected graft explantation and graft preservation. CI, Confidence interval.
of the aorta with a synthetic graft. Currently, in spite of the lack of prospective randomized trials, TEVAR is accepted as the treatment of choice for most cases of thoracic aortic disease that have anatomic feasibility; exceptions might apply mainly to patients with connective tissue diseases. Therefore, a relevant portion of the most recent data available on thoracic aortic graft infection actually refers to EG. Moulakakis et al48 in 2014 reviewed the literature for graft infection after TEVAR. They analyzed 55 patients treated with preservation of the EG and 41 with excision. Preservation of the EG with antibiotic therapy alone was shown not to be a durable option; their meta-analysis showed a trend for better outcomes in patients with EG excision. The authors found a trend for worse outcomes in patients with fistulas. Moreover, there was no difference whether the patients were treated with in situ or extra-anatomic reconstructions.
Our review identified 43 articles reporting on 49 patients with thoracic aortic SG infection and 184 patients with thoracic EG infection. Among these, four were multicenter studies, all focused on EG infection, associated with a fistula in 91% of cases. Patients presented with hematemesis or hemoptysis in 53% to 92% of cases and shock in 22% to 26%. Management included several different strategies that we may divide into “purely conservative” (19%-42% of cases) and somewhat “operative” (58%-81%). The operative cases included open aortic replacement, with or without associated esophageal or bronchial repair; repeated TEVAR, with or without associated esophageal or bronchial repair; and esophageal or bronchial open repair or endoscopic treatment (stenting) alone. Outcome reporting was totally heterogeneous in these multicenter reports, mainly represented by survival at different time points, and not specifying results per single management
Journal of Vascular Surgery Volume
-,
Number
Kahlberg et al
9
-
strategy in most cases. Nevertheless, conservative treatment was associated with the poorest survival (0%-20% at 1 year), whereas operative treatment showed a survival ranging from 27% to 39% at different follow-up points (up to 5 years). Among patients described in the 39 single-center studies, the time interval between index procedure and diagnosis of infection was shorter in patients with previous endografting than in patients with previous surgical repair. This result may be explained by the typically different cohort of patients treated with TEVAR compared with surgically treated ones. In particular, TEVAR is usually the preferred strategy for treatment of aortic pseudoaneurysms and saccular aneurysms, especially in an emergency, and a proportion of these cases may represent “undiagnosed” infected aneurysms or pseudoaneurysms. Moreover, the enhanced medical surveillance that is usually advised after TEVAR may account for a significant reduction in the time needed to diagnose infectious complications. Patients with EG infection were also more frequently reported to have an associated AEF than patients with SG infection. Even if detailed causes of this process are still unclear, this finding is consistent with a number of studies confirming that both thoracic and abdominal aortic endografting may be complicated by gastrointestinal fistulization, at least at the same rate as after surgical treatment.5,7,49 This review of single-center published series and case reports confirmed that clinical outcomes of thoracic aortic prosthetic infections are poor, with overall survival of 45% at 1 year and 19% at 5 years. Probably because of the small number of cases, meta-analyses failed to reach statistical significance in showing an undisputable difference in 1-year mortality between EG infection and SG infection. However, a slight (nonsignificant) trend was noted toward worse outcomes in case of SG infection that may be explained by the higher surgical complexity of “redo” interventions in case of previous open aortic surgery compared with patients who experienced only endovascular interventions, with presence of more extensive adhesions and “frozen” chest. Similarly, in comparing infected graft explantation vs graft preservation, the graft explantation strategy seemed to be associated with a better outcome, even if this result may be biased by the small number of studies included and by the low statistical power. Nonetheless, this finding is consistent with the previous published evidence on superiority of EG explantation vs preservation in patients with post-TEVAR infections,48 probably reflecting the inability of antibiotic therapy alone to eradicate infection involving either a surgical or an endovascular aortic prosthetic graft. We acknowledge some limitations of this review. The heterogeneity of the included patients (infection of surgical or endovascular graft, presence or absence of
associated fistula, conservative or surgical management) makes it difficult to draw consistent conclusions from pooling to drive clinical decision-making. However, consideration should be given to the results presented because they are at present the most comprehensive data on a topic that is still not sufficiently addressed in the literature. The decision to include a large number of small studies, including case reports, is driven by the rarity of this disease. This may account for publication bias due to the plausible attitude of small centers to report most “interesting” and “successful” cases. Also, a time bias should be considered in comparing results of SG infection with the more recent cases of EG infection. Furthermore, a number of studies were excluded for lacking or incomplete reporting.
CONCLUSIONS This systematic review showed that thoracic aortic EG infection presents some differences compared with SG infection regarding frequency of associated aortoesophageal fistulization, time interval between index procedure and diagnosis of infection, and symptoms at presentation. Overall survival is poor in both groups, but meta-analysis found a trend of worse outcome in patients with SG infection. Even if treatment strategies are various and depend on each center’s experience and on specific patients’ conditions, the current evidence probably seems to support surgical treatment with graft explantation when it is feasible. However, these data are so far too weak to justify changes in clinical practice, and more comparative studies are needed to provide clinical recommendations in the future.
AUTHOR CONTRIBUTIONS Conception and design: AK, GM, RC Analysis and interpretation: AK, AG, FV, NM, NC, GM Data collection: AK, AG, DL, GM Writing the article: AK, AG, DL, GM Critical revision of the article: AK, FV, NM, NC, GM, RC Final approval of the article: AK, AG, DL, FV, NM, NC, GM, RC Statistical analysis: AK, AG Obtained funding: Not applicable Overall responsibility: GM
REFERENCES 1. Kahlberg A, Melissano G, Mascia D, Loschi D, Grandi A, Chiesa R. How to best treat infectious complications of open and endovascular thoracic aortic repairs. Semin Vasc Surg 2017;30:95-102. 2. Moher D, Liberati A, Tetzlaff J, Altman DG; PRISMA Group. Preferred reporting items for systematic reviews and metaanalyses: the PRISMA statement. BMJ 2009;339:b2535. 3. Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 2000;283:2008-12.
10
Kahlberg et al
Journal of Vascular Surgery ---
4. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003;327:557-60. 5. Chiesa R, Melissano G, Marone EM, Marrocco-Trischitta MM, Kahlberg A. Aorto-oesophageal and aortobronchial fistulae following thoracic endovascular aortic repair: a national survey. Eur J Vasc Endovasc Surg 2010;39:273-9. 6. Smeds MR, Duncan AA, Harlander-Locke MP, Lawrence PF, Lyden S, Fatima J, et al. Treatment and outcomes of aortic endograft infection. J Vasc Surg 2016;63:332-40. 7. Czerny M, Eggebrecht H, Sodeck G, Weigang E, Livi U, Verzini F, et al. New insights regarding the incidence, presentation and treatment options of aorto-oesophageal fistulation after thoracic endovascular aortic repair: the European Registry of Endovascular Aortic Repair Complications. Eur J Cardiothorac Surg 2014;45:452-7. 8. Czerny M, Reser D, Eggebrecht H, Janata K, Sodeck G, Etz C, et al. Aorto-bronchial and aorto-pulmonary fistulation after thoracic endovascular aortic repair: an analysis from the European Registry of Endovascular Aortic Repair Complications. Eur J Cardiothorac Surg 2015;48:252-7. 9. Kieffer E, Sabatier J, Plissonnier D, Knosalla C. Prosthetic graft infection after descending thoracic/thoracoabdominal aortic aneurysmectomy: management with in situ arterial allografts. J Vasc Surg 2001;33:671-8. 10. Kieffer E, Chiche L, Gomes D. Aortoesophageal fistulae: value of in situ aortic allograft replacement. Ann Surg 2003;238: 283-90. 11. Eggebrecht H, Mehta RH, Dechene A, Tsagakis K, Kühl H, Huptas S, et al. Aortoesophageal fistula after thoracic aortic stent-graft placement. A rare but catastrophic complication of a novel emerging technique. JACC Cardiovasc Interv 2009;2:570-6. 12. Czerny M, von Allmen R, Opfermann P, Sodeck G, Dick F, Stellmes A, et al. Self-made pericardial tube graft: a new surgical concept for treatment of graft infections after thoracic and abdominal aortic procedures. Ann Thorac Surg 2011;92:1657-62. 13. Lemaire SA, Green SY, Kim JH, Sameri A, Parenti JL, Lin PH, et al. Thoracic or thoracoabdominal approaches to endovascular device removal and open aortic repair. Ann Thorac Surg 2012;93:726-33. 14. Lyons OT, Patel AS, Saha P, Clough RE, Price N, Taylor PR. A 14-year experience with aortic endograft infection: management and results. Eur J Vasc Endovasc Surg 2013;46: 306-13. 15. Canaud L, Alric P, Gandet T, Ozdemir BA, Albat B, MartyAne C. Open surgical secondary procedures after thoracic endovascular aortic repair. Eur J Vasc Endovasc Surg 2013;46: 667-74. 16. Kawamoto S, Sato M, Motoyoshi N, Kumagai K, Adachi O, Saito T, et al. Outcomes of a staged surgical treatment strategy for aortoesophageal fistula. Gen Thorac Cardiovasc Surg 2015;63:147-52. 17. Luehr M, Etz CD, Nozdrzykowski M, Garbade J, Lehmkuhl L, Schmidt A, et al. Emergency open surgery for aortooesophageal and aorto-bronchial fistulae after thoracic endovascular aortic repair: a single-centre experience. Eur J Cardiothorac Surg 2014;47:374-83. 18. Suzuki T, Kawamoto S, Motoyoshi N, Akiyama M, Kumagai K, Adachi O, et al. Contemporary outcome of the surgical management of prosthetic graft infection after a thoracic aortic replacement: is there a room to consider vacuumassisted wound closure as an alternative? Gen Thorac Cardiovasc Surg 2014;63:86-92. 19. Kahlberg A, Melissano G, Tshomba Y, Leopardi M, Chiesa R. Strategies to treat thoracic aortitis and infected aortic grafts. J Cardiovasc Surg (Torino) 2015;56:269-80.
2018
20. Coselli JS, Koksoy C, LeMaire SA. Management of thoracic aortic graft infections. Ann Thorac Surg 1999;67:1990-8. 21. Fatima J, Duncan AA, De Grandis E, Oderich GS, Kalra M, Gloviczki P, et al. Treatment strategies and outcomes in patients with infected aortic endografts. J Vasc Surg 2013;58: 371-9. 22. Kubota H, Endo H, Noma M, Ishii H, Tsuchiya H, Yoshimoto A, et al. Xenopericardial roll graft replacement for infectious pseudoaneurysms and graft infections of the aorta. J Cardiothorac Surg 2015;5:1-9. 23. VonFricken K, Karamanoukian HL, Ricci M, Taheri A, Bergsland J, Salerno TA. Aortobronchial fistula after endovascular stent graft repair of the thoracic aorta. Ann Thorac Surg 2000;70:1407-9. 24. Bond SE, McGuinness CL, Reidy JF, Taylor PR. Repair of secondary aortoesophageal fistula by endoluminal stentgrafting. J Endovasc Ther 2001;8:597-601. 25. Hance KA, Hsu J, Eskew T, Hermreck AS. Secondary aortoesophageal fistula after endoluminal exclusion because of thoracic aortic transection. J Vasc Surg 2003;37:886-8. 26. Porcu P, Chavanon O, Sessa C, Thony F, Aubert A, Blin D. Esophageal fistula after endovascular treatment in a type B aortic dissection of the descending thoracic aorta. J Vasc Surg 2005;41:708-11. 27. Abdul-Ghani A, Pisipati S, McWilliams R, Page RD. Aortobronchial fistula following aortic and bronchial stenting of a thoracic aneurysm. Eur J Cardiothorac Surg 2006;29:419-21. 28. Riesenman PJ, Farber MA, Mauro MA, Selzman CH, Feins RH. Aortoesophageal fistula after thoracic endovascular aortic repair and transthoracic embolization. J Vasc Surg 2007;46: 789-91. 29. Martens K, De Mey J, Everaert H, Delvaux G, van den Brande P. Aortoesophageal fistula following endovascular exclusion of a thoracic aneurysm. Int Angiol 2007;26: 292-6. 30. Civilini E, Bertoglio L, Melissano G, Chiesa R. Aortic and esophageal endografting for secondary aortoenteric fistula. Eur J Vasc Endovasc Surg 2008;36:297-9. 31. Isasti G, Gómez-Doblas JJ, Olalla E. Aortoesophageal fistula: an uncommon complication after stent-graft repair of an aortic thoracic aneurysm. Interact Cardiovasc Thorac Surg 2009;9:683-4. 32. Riesenman PJ, Farber MA. Management of a thoracic endograft infection through an ascending to descending extra-anatomic aortic bypass and endograft explantation. J Vasc Surg 2010;51:207-9. 33. Yavuz S, Kanko M, Ciftci E, Parlar H, Agirbas H, Berki T. Aortoesophageal fistula secondary to thoracic endovascular aortic repair of a descending aortic aneurysm rupture. Heart Surg Forum 2011;14:E249-51. 34. Lee SH, Song PS, Kim WS, Park KB, Choi S. A case of stent graft infection coupled with aorto-esophageal fistula following thoracic endovascular aortic repair in a complex patient. Korean Circ J 2012;42:366-8. 35. Yamashita S, Dohi T, Shimizu Y, Momomura S. Aortopulmonary fistula caused by an infected thoracic aortic false aneurysm rupturing after endovascular stent placement. Drug Discov Ther 2012;6:278-82. 36. Okamoto H, Tamenishi A, Matsumura Y, Niimi T. Composite vein graft reconstruction for infected descending aortic prosthesis. Ann Thorac Surg 2012;93:2061-3. 37. Akkoyunlu Y, Aslan T, Ceylan B, Inan B, Aydin C, Arvas G. Aortic stent infection in etiology of fever of unknown origin in a geriatric patient. Asian Pac J Trop Dis 2012;2(Suppl 2): S963-4. 38. Onodera M, Inoue Y, Fujino Y, Kikuchi S, Endo S. A case of secondary aortoesophageal fistula inserted a covered
Journal of Vascular Surgery Volume
39.
40.
41.
42.
43.
44.
-,
Number
Kahlberg et al
11
-
self-expanding esophageal stent to control gastrointestinal bleeding. Case Rep Gastrointest Med 2013;2013:857135. Muradi A, Yamaguchi M, Kitagawa A, Nomura Y, Okada T, Okita Y, et al. Secondary aortoesophageal fistula after thoracic endovascular aortic repair for a huge aneurysm. Diagn Interv Radiol 2013;19:81-4. M, Bioc ina B, Uzun S, Mestrovic T, Safradin Petrunic I. Cryopreserved aortic homograft for in situ replacement of infected thoracic stent graft associated with distal aortic arch rupture and hematemesis. Ann Thorac Surg 2014;98: 2219-21. Yamasaki M, Abe K, Misumi H, Ito J, Nakanishi Y. Endograft infection after hybrid surgery for chronic Stanford type B aortic dissection: endograft infection and treatment. Surg Today 2015;45:1575-8. Georvasili VK, Bali C, Peroulis M, Kouvelos G. Management of an aorto-esophageal fistula, complicating a descending thoracic aortic aneurysm endovascularly repaired. Gen Thorac Cardiovasc Surg 2016;64:216-9. Sueda T, Takahashi S, Katayama K. Successful treatment of an infected thoracic endovascular stent graft. Gen Thorac Cardiovasc Surg 2016;64:273-6. Afifi RO, Mushtaq HH, Sandhu HK, Khalil K, Safi HJ, Estrera AL. Successful multistaged surgical management of secondary aortoesophageal fistula with graft infection. Ann Thorac Surg 2016;101:e203-5.
45. Yamamoto M, Nishimori H, Iguchi M, Orihashi K. Pathological investigation of graft-related oesophageal fistula. Interact Cardiovasc Thorac Surg 2017;24:813-4. 46. Clarke NS, Murthy R, Reznik SI, Jessen ME. Coxiella burnettiassociated thoracic endovascular stent graft infection. J Card Surg 2017;32:506-7. 47. Sladojevic M, Bjelovic M, Ilic N, Mutavdzic P, Koncar I, Dragas M, et al. Open surgical treatment of secondary aortoesophageal and aortobronchial fistula after thoracic endovascular aortic repair and esophagocoloplasty in a second procedure. Ann Vasc Surg 2017;44:417.e11-6. 48. Moulakakis KG, Mylonas SN, Antonopoulos CN, Kakisis JD, Sfyroeras GS, Mantas G, et al. Comparison of treatment strategies for thoracic endograft infection. J Vasc Surg 2014;60:1061-71. 49. Kahlberg A, Rinaldi E, Piffaretti G, Speziale F, Trimarchi S, Bonardelli S, et al; MAEFISTO collaborators. Results from the Multicenter Study on Aortoenteric Fistulization After Stent Grafting of the Abdominal Aorta (MAEFISTO). J Vasc Surg 2016;64:313-20.e1. Submitted May 1, 2018; accepted Oct 26, 2018.
Additional material for this article may be found online at www.jvascsurg.org.
11.e1
Journal of Vascular Surgery
Kahlberg et al
---
2018
Supplementary Table (online only). Quality assessment of included studies with the Newcastle-Ottawa Scale Authors
Year
Selection (of 4)
Comparability (of 2)
Outcome (of 3)
Total No. of stars (of 9)
Study quality
VonFricken
2000
3
1
3
7
High
Bond
2001
3
1
3
7
High
Hance
2003
3
2
3
8
High
Porcu
2005
2
1
3
6
Moderate
Abdul-Ghani
2006
2
2
3
7
High
Riesenman
2007
2
2
3
7
High
Martens
2007
3
1
3
7
High
Civilini
2008
3
1
3
7
High
Isasti
2009
3
1
3
7
High
Riesenman
2010
2
1
3
6
Moderate
Yavuz
2011
3
1
3
7
High
Lee
2012
3
1
3
7
High
Yamashita
2012
3
1
3
7
High
Okamoto
2012
3
1
3
7
High
Akkoyunlu
2012
3
2
3
8
High
Onodera
2013
3
1
3
7
High
Muradi
2013
3
2
3
8
High
Petrunic
2014
3
1
3
7
High
Yamasaki
2015
3
1
3
7
High
Georvasili
2016
3
1
3
7
High
Sueda
2016
3
1
3
7
High
Afifi
2016
3
1
3
7
High
Yamamoto
2017
3
1
3
7
High
Clarke
2017
3
1
1
5
Moderate
Sladojevic
2017
3
1
3
7
High
Coselli
1999
4
2
3
9
High
Fatima
2013
4
2
3
9
High
Kubota
2015
4
2
3
9
High
Kieffer
2001
4
2
3
9
High
Kieffer
2003
4
2
3
9
High
Eggebrecht
2009
4
2
3
9
High
Czerny
2011
4
2
1
7
High
LeMaire
2012
4
2
3
9
High
Lyons
2013
4
2
3
9
High
Canaud
2013
4
2
3
9
High
Kawamoto
2014
4
2
3
9
High
Luehr
2014
4
2
3
9
High
Suzuki
2014
4
2
3
9
High
Kahlberg
2015
4
2
3
9
High
Chiesa
2009
4
2
3
9
High
Czerny
2014
4
2
3
9
High
Czerny
2015
4
2
3
9
High
Smeds
2016
4
2
3
9
High
ABSTRACT Objective: The objective of this study was to collect and critically analyze the current evidence on the modalities and results of treatment of descending thoracic aortic surgical graft (SG) and endograft (EG) infection, which represents a rare but dramatic complication after both surgical and endovascular aortic repair. Methods: A comprehensive electronic health database search (PubMed/MEDLINE, Scopus, Google Scholar, and the Cochrane Library) identified all articles that were published up to October 2017 reporting on thoracic aortic SG or EG infection. Observational studies, multicenter reports, single-center series and case reports, case-control studies, and guidelines were considered eligible if reporting specific results of treatment of descending thoracic aortic SG or EG infection. Comparisons of patients presenting with SG or EG infection and between invasive and conservative treatment were performed. Odds ratio (OR) meta-analyses were run when comparative data were available. Results: Forty-three studies reporting on 233 patients with infected SG (49) or EG (184) were included. Four were multicenter studies including 107 patients, all with EG infection, associated with a fistula in 91% of cases, with a reported overall survival at 2 years of 16% to 39%. The remaining 39 single-center studies included 49 patients with SG infection and 77 with EG infection. Association with aortoesophageal fistula was significantly more common with EG (60% vs 31%; P ¼ .01). In addition, time interval from index procedure to infection was significantly shorter with EG (17 6 21 months vs 32 6 61 months; P ¼ .03). Meta-analysis showed a trend of increased 1-year mortality in patients with SG infection compared with EG infection (pooled OR, 3.6; 95% confidence interval, 0.9-14.7; P ¼ .073). Surgical management with infected graft explantation was associated with a trend toward lower 1-year mortality compared with graft preservation (pooled OR, 0.3; 95% confidence interval, 0.1-1.0; P ¼ .056). Conclusions: Thoracic aortic EG infection is likely to occur more frequently in association with aortoesophageal fistulas and in a shorter time compared with SG infection. Survival is poor in both groups, especially in patients with SG infection. Surgical treatment with graft explantation seems to be the preferable choice in fit patients. (J Vasc Surg 2018;-:1-11.) Keywords: Infection; Blood vessel prosthesis; Thoracic aorta; Surgery; Endovascular procedures; Fistula; Esophageal fistula; Respiratory fistula
Vascular prosthetic infection is a serious and often fatal event; moreover, infection of an aortic surgical graft (SG) or endograft (EG) previously implanted at the descending thoracic aortic level is considered one of the most feared complications of modern vascular surgery.1 In the last decades, the number of thoracic aortic proceduresdespecially endovasculardis continuously increasing, and infectious prosthetic complications
From the Department of Vascular Surgery, Vita-Salute University School of Medicine, San Raffaele Scientific Institute, Milana; the Department of Vascular and Thoracic Surgery, Ghent University Hospital, Ghentb; and the Department
have been observed more and more frequently, affecting 1% to 6 % of patients and entailing up to 75% mortality rates even when recognized and treated.1 Nevertheless, the evidence is poor and consensus documents are lacking in regard to the incidence, the clinical features, the therapeutic options, and above all the clinical outcome of this disease. The aim of this study was to collect and critically analyze the published evidence on the management and outcome of treatment of prosthetic infection after previous repair of the descending thoracic aorta to compare mortality of patients presenting with SG or EG infection and of patients managed conservatively or surgically.
of Vascular Surgery and Kidney Transplantation, University Hospital of Strasbourg, Strasbourg.c Author conflict of interest: none. Additional material for this article may be found online at www.jvascsurg.org. Correspondence: Andrea Kahlberg, MD, Department of Vascular Surgery, San Raffaele Scientific Institute, Vita-Salute University School of Medicine, Via Olgettina 60, 20132 Milan, Italy (e-mail: [email protected]). The editors and reviewers of this article have no relevant financial relationships to disclose per the JVS policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest. 0741-5214 Copyright Ó 2018 by the Society for Vascular Surgery. Published by Elsevier Inc. https://doi.org/10.1016/j.jvs.2018.10.108
METHODS Study design and search strategies. The review conformed to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement standards.2 An extensive electronic health database search was undertaken to identify all the articles that were published before November 2017 reporting on SG and EG infections after surgical repair or endovascular repair of thoracic or thoracoabdominal aortic disease. 1
2
Kahlberg et al
Journal of Vascular Surgery ---
The following electronic bibliographic sources were searched: PubMed/MEDLINE, Scopus, Google Scholar, and the Cochrane Library. Search strategies included MESH terms and key words: (“aorta, thoracic”[MeSH Terms] OR (“aorta”[All Fields] AND “thoracic”[All Fields]) OR “thoracic aorta”[All Fields]) AND (“infection”[MeSH Terms] OR “infection”[All Fields]) OR (“thoracic”[All Fields] AND “Stent-Graft”[All Fields]) AND (“infection”[MeSH Terms] OR “infection”[All Fields]) OR (“thoracic”[All Fields] AND “aorta”[All Fields]) OR “graft”[All Fields] OR “Stent-Graft”[All Fields] AND “infection”[All Fields] OR (“graft”[All Fields] AND “infection”[All Fields]) OR “aortoesophageal”[All Fields] AND (“fistula”[MeSH Terms] OR “fistula”[All Fields]) OR “aorto-oesophageal”[All Fields] AND (“fistula”[MeSH Terms] OR “fistula”[All Fields]) OR “aortobronchial”[All Fields] AND (“fistula”[MeSH Terms] OR “fistula”[All Fields]). In addition, the reference lists of all retrieved articles were examined for further relevant studies. The last search was run in October 2017. Articles in languages other than English were excluded. Eligibility criteria and study selection. Initial study selection and systematic review inclusion and exclusion were performed by two authors (A.G., D.L.) on the basis of the following eligibility criteria. Observational studies, multicenter reports, single-center series and case reports, case-control studies, and guidelines were included in the review analysis. No randomized clinical trial was found covering this topic. Articles reporting on descending thoracic EG infection in patients with initial associated supra-aortic trunk debranching (ie, proximal landing zone 1 or zone 2) were included. Articles reporting only on primary aortic infections (eg, mycotic aneurysm, infectious aortitis, primary aortoesophageal fistula [AEF] or aortobronchial fistula [ABF]), even if affected patients were treated with surgical or endovascular repair, were excluded. Papers limited to graft infection of the ascending aorta, of the arch, or of the infrarenal aorta and papers expressing general principles of vascular graft infection did not meet our eligibility criteria. Articles failing to provide mortality data specifically related to patients with descending thoracic or thoracoabdominal aortic graft infection (eg, reporting only about diagnostic modalities or mixing results of abdominal and thoracic infections) were excluded. When multiple publications on the same patient sample were identified or study populations overlapped, only the latest report was included unless the reported outcomes were mutually exclusive. Data extraction and study quality assessment. Each study (full-text article) was reviewed by two independent reviewers (A.K., A.G.), and the following data were extracted and inserted in a precompiled electronic database sheet: number of patients, baseline demographics,
2018
type of index procedure (surgical or endovascular), indication for index procedure, time interval to infection diagnosis, clinical presentation, type of management, perioperative mortality, and mortality at follow-up. If necessary, the article’s corresponding author was contacted to confirm or to clarify specific information. The quality of evidence was assessed at the individual study level by two authors (A.K., D.L.; Supplementary Table, online only). Summary process and results reporting. Because of the possible overlap of populations of patients with single-center reports, analysis of the results of multicenter studies was performed separately, reporting results (including follow-up survival) as described in the original articles and avoiding summary measures and comparative analysis. For single-center reports, results were reported summarizing (mean and proportions) available data extracted by each study, according to the type of infected graft (SG vs EG). Pooled mortality rate was reported at 30 days, 1 year, and 5 years. Meta-analysis was performed including only observational studies reporting results of direct comparison between given subgroups of patients (see Statistical Analysis). For all meta-analyses, the primary end point was considered the overall mortality rate at 1 year. Statistical analysis. Standard descriptive statistics were reported as numbers and percentages (categorical variables) or mean and range (continuous variables). Variables’ distribution was tested for normality by using the D’Agostino-Pearson test. Comparisons between patients presenting with SG infection or with EG infection were performed using two-tailed t-test for normally distributed continuous variables, Mann-Whitney test for non-normally distributed continuous variables, and Fisher exact test for categorical variables. Three odds ratio (OR) meta-analyses of all eligible series in accordance with the recommendations of the Metaanalysis of Observational Studies in Epidemiology group3 were performed: SG infection vs EG infection; conservative medical treatment vs surgical (any kind) treatment; and infected graft explantation vs preservation (considering only surgically treated patients). Pooled ORs were initially calculated using a fixed-effect model and reported as forest plots. Heterogeneity of included studies was assessed using the I2 statistics.4 When heterogeneity was considered high, ORs calculated under the random-effects model were preferred. Standard descriptive and comparison statistics were run using Prism 7 statistical software (GraphPad Software, Inc, La Jolla, Calif). The meta-analyses were performed with MedCalc statistical software, version 17.9.7 (MedCalc Software, Ostend, Belgium).
Journal of Vascular Surgery Volume
-,
Number
Kahlberg et al
3
-
Fig 1. Flow chart of studies identified and finally included in the review analysis.
RESULTS The initial literature search identified 1657 records, of which 239 were duplicates, and another 1200 were considered irrelevant after exclusion criteria application at the title or abstract level. Among the remaining 218 articles assessed for eligibility, 175 were excluded because of their study design (including 4 systematic reviews/metaanalyses summarizing the results of papers already individually evaluated and 10 papers reporting only results of diagnostic imaging) or because they reported insufficient data on study outcomes. Finally, 43 studies (the first one published in June 1999 and the last one in October 2017) reporting on 233 patients with infected SG (12 studies, 49 patients) or infected EG (31 studies, 184 patients) were included in the qualitative and quantitative analysis (Fig 1). Multicenter reports. Among included articles, four were based on large, multicenter retrospective series5-8 including 107 patients. All of these reported only data of patients with thoracic EG infection. No articles
providing multicenter reports on thoracic SG infection were found in the literature. Analysis of results of multicenter studies is summarized in Table I. Three studies reported the incidence and outcome of AEF and ABF after thoracic endovascular aortic repair (TEVAR)5,7,8; the fourth study reported the incidence and outcome of EG infection after both endovascular aneurysm repair (EVAR) and TEVAR.6 The mean age of considered patients was 70 years, and the male-female ratio was 3:1. The indication for primary TEVAR, when reported, was an atherosclerotic aneurysm in 69% of cases. The type of infected EG was reported in 60 of 107 cases, of which 42 had polyethylene terephthalate (PET)-based fabric and 18 had expanded polytetrafluoroethylene-based fabric. Specific graft models were Valiant in 2 cases (Medtronic, Santa Rosa, Calif), Talent in 11 cases (Medtronic), TAG in 18 cases (W. L. Gore & Associates, Flagstaff, Ariz), Zenith TX1 in 2 cases (Cook Medical, Bloomington, Ind), Zenith TX2 in 8 cases (Cook Medical), Endofit in 3 cases (Endomed, Phoenix, Ariz), Relay in 5 cases (Bolton
4
Journal of Vascular Surgery
Kahlberg et al
---
2018
Table I. Demographics, indication for index procedure, timing of infection diagnosis, clinical presentation, management strategies, and outcome of 107 patients described in multicenter case series of infections after thoracic endovascular aortic repair (TEVAR) Chiesa et al,5 2010
Smeds et al,6 2016
Czerny et al,7 2014
Czerny et al,8 2015
Total
19
26
36
26
107
Age, years
73.8 6 7.1
68
69 (56-75)
70 (60-77)
Male
16 (84)
No. of patients
20 (77)
27 (75)
17 (65)
80 (75)
28 (77)
Indication for index procedure Atherosclerotic aneurysm
13 (68)
e
Dissection, PAU, IMH
2 (11)
e
Traumatic aortic injury
1 (5)
e
e
e
Mycotic, AEF, ABF Secondary pseudoaneurysm Other
3 (16) e
Timing from index procedure to diagnosis of infection, days
327 (e)
15 (58)
56 (69)
e
6 (23)
8 (18)
e
4 (15)
5 (11)
5 (14)
e
5 (9)
e
e
1 (4)
4 (9)
e
3 (8)
e
3 (8)
540 (5-2100)
90 (30-150)
310 (28-1065)
317 (5-2100)
Clinical presentation Pain
1 (5)
17 (66)
Fever or chills
11 (58)
17 (66)
Hematemesis or hemoptysis
13 (68) 5 (26)
Shock Dyspnea
2 (10)
AEF
4 (15)
22 (31)
29 (81)
7 (27)
64 (60)
e
19 (53)
24 (92)
56 (69)
e
8 (22)
6 (23)
19 (23)
e
2 (5)
e
13 (68)
12 (46)
ABF
5 (26)
0 (0)
AEF þ ABF
5 (26)
0 (0)
Conservative
8 (42)
5 (19)
Fistula repair only
6 (32)
e
Surgical þ fistula repair
1 (5)
e
Surgical fistula repair
1 (5)
7 (27)
11 (14)
0 (0)
61 (57)
0 (0)
26 (100)
31 (29)
0 (0)
0 (0)
5 (5)
10 (28)
5 (19)
28 (26)
13 (36)
4 (15)
23 (28)
13 (36)
8 (31)
22 (27)
e
e
22 (49)
36 (100)
a
Management
21 (81)
Endo þ fistula repair
1 (5)
e
e
2 (8)
3 (7)
Endo fistula repair
2 (11)
e
e
7 (27)
9 (20)
28% (1-year)
39% (2-year)
Outcome (survival) Overall
16% (2-year)
29% (5-year)
Conservative
0% (30-day)
20% (1-year)
0% (1-year)
e
Stent, esophageal only
e
e
17% (1-year)
e
Open, esophageal only
e
e
43% (1-year)
e
Open þ esophageal
e
e
46% (1-year)
e
27% (2-year)
29% (5-year)
e
e
Open
ABF, Aortobronchial fistula; AEF, aortoesophageal fistula; IMH, intramural hematoma; PAU, penetrating aortic ulcer. Categorical variables are presented as number (%). Continuous variables are presented as mean 6 standard deviation or median (interquartile range). a Conservative, medical treatment with or without minimally invasive additional procedures; Fistula repair only, esophageal or respiratory fistula repair, without associated aortic repair; Surgical þ fistula repair, surgical open aortic repair associated with additional maneuvers to repair the fistula; Surgical fistula repair, surgical open aortic repair without additional maneuvers to repair the fistula; Endo þ fistula repair, endovascular aortic repair (TEVAR) associated with additional maneuvers to repair the fistula; Endo fistula repair, endovascular aortic repair (TEVAR) without additional maneuvers to repair the fistula.
Medical, Sunrise, Fla), and Powerlink in 1 case (Endologix, Irvine, Calif). In regard to possible etiologic factors of subsequent EG infection, only Smeds et al6 (grouping together abdominal and thoracic EVARs) mentioned an infectious complication during or after the index operation occurring in 34% of cases (urinary tract infection in 8%, groin infection in 7%,
other unspecified infection in 19%). The same group reported the occurrence of non-aorta-related secondary procedures between the index operation and the diagnosis of infection in 69 patients (34%). The mean time interval from index procedure to diagnosis of infection was 295 days (approximately 10 months). In 91% of reported patients, a fistula was
Journal of Vascular Surgery Volume
-,
Number
Kahlberg et al
5
-
present at diagnosis. Imaging modalities used to make or to confirm the diagnosis of aortic graft infection were computed tomography scan in 80 cases and magnetic resonance imaging in 1 case. In case of associated fistula, the diagnosis was made by means of esophagogastroduodenoscopy in 45 cases, aortography in 2 cases, and esophageal transit study in 3 cases. Only the study by Smeds et al6 specifically reported the results of bacteriologic findings. Perioperative blood cultures were positive in 63% of patients; intraoperative cultures identified gram-positive organisms in 22%, gram-negative organisms in 13%, and fungus in 5%. Intraoperative cultures were polymicrobial in 35% (and this was also considered a significant risk factor for morbidity and mortality in patients managed surgically) and negative in 30% of cases. A number of different strategies to treat EG infection were reported (Table I). Smeds et al6 reported the type of graft used for aortic reconstruction, as follows: a prosthetic graft in 16 cases (83% antibiotic soaked), a cryopreserved allograft in 3 cases, and a femoropopliteal neoaortoiliac system in 2 cases. Reported postoperative complications after surgical management of thoracic aortic graft infection included respiratory failure in 15 patients, bleeding in 13, multiorgan failure in 8, renal dysfunction in 5, mediastinitis in 4, and acute myocardial infarction in 3. At a mean follow-up of 25 months (range, 1-60 months), overall survival was 28% (30/107). Open aortic replacement showed a survival of 27% at 2 years in one study and of 29% at 5 years in another study. Surgical esophageal repair showed a survival of 43% at 1 year when performed alone and a survival of 46% at 1 year when associated with open aortic repair. Conservative strategies were generally associated with worse outcomes, showing a survival of 0% at 30 days in one study, 0% at 1 year in another study, and 20% at 1 year in the last study. Similarly, esophageal stenting alone showed a survival of 17% at 1 year. Single-center reports. The remaining 39 articles reported single-center experience: 11 were case series with >5 patients,9-19 3 were case series with #5 patients,20-22 and 25 were case reports,23-47 with a total of 126 patients included in this analysis. Among these, 49 patients presented with thoracic aortic SG infection (41 descending thoracic grafts and 8 thoracoabdominal grafts), and 77 presented with thoracic aortic EG infection. Among EG infection cases, the type of infected EG was reported in 28 of 77 cases, of which 20 had PET-based fabric and 8 had expanded polytetrafluoroethylene-based fabric. Specific graft models were Valiant in 9 cases (Medtronic), Talent in 7 cases (Medtronic), TAG in 7 cases (W. L. Gore), Zenith TX2 in 2 cases (Cook Medical), Seal thoracic stent graft in 1 case (S&G Biotech Inc, Seoul, Korea), GianturcoRosch Z stent in 1 case (assembled at authors’ institution),
and Bard custom-made thoracic stent graft in 1 case (C.R. Bard, Tempe, AZ). Demographics, data on clinical presentation, and outcomes of the two groups are presented in Table II. Eight papers mentioned some infectious complications occurring during the perioperative period of the index procedure. In particular, 7 papers9,27,28,37,39,40,45 (including a total population of 17 patients) reported pulmonary infection occurring in 9 patients, septicemia in 2, urinary tract infection in 1, coagulopathy in 1, and early duodenalparaprosthetic fistula in 1 open thoracoabdominal repair case. Another paper,14 reporting on 22 patients with EG infection (but mixing abdominal and thoracic EVARs), described 11 patients (50%) experiencing perioperative septic complications. Secondary vascular procedures occurring between the index procedure and the diagnosis of infection were reported in five cases, including one TEVAR after open thoracoabdominal aortic repair due to proximal disease progression, two cases of embolization of a type I endoleak, one bronchial stenting due to aneurysm sac compression after TEVAR, and one emergent visceral surgical debranching for thoracoabdominal disease progression. Nonvascular procedures were reported in five patients, including two cases of pulmonary resection associated with aortic replacement, one case of belowknee amputation during the interval, one cholecystectomy, and one case of multiple hospitalizations requiring placement of central venous catheters. In SG, the mean time interval from index procedure to diagnosis of infection was 32.2 months, significantly longer compared with EG (17.1 months; P ¼ .03). Association with AEF was significantly less common with SG than with EG (31% vs 60%, respectively; P ¼ .01). Symptoms at presentation, pain and fever or chills, were significantly more common with SG (73% vs 35% [P < .001] and 84% vs 61% [P ¼ .02], respectively), whereas hematemesis was significantly more common with EG (42% vs 19%; P ¼ .02). Imaging modalities used to make or to confirm the original diagnosis of aortic graft infection were computed tomography scan in 51 cases, fluorodeoxyglucose positron emission tomography in 3 cases, and magnetic resonance imaging in 1 case. In case of associated fistula, the diagnosis was made by means of esophagogastroduodenoscopy in 29 cases, aortography in 5 cases, and esophageal transit study in 4 cases. Specific antimicrobial findings were reported in 18 papers (53 isolated microorganisms in 49 patients). In almost all cases, findings were based on blood cultures. Most commonly isolated micro-organisms were Staphylococcus aureus (19 patients, methicillin resistant [MRSA] in 7) and Streptococcus spp (12 patients); other isolated microorganisms were Escherichia coli (5), Pseudomonas aeruginosa (4), Candida albicans (4), Enterococcus spp (3), Klebsiella spp (2), Lactobacillus spp (1), Haemophilus influenzae (1), Serratia spp (1), and
6
Journal of Vascular Surgery
Kahlberg et al
---
2018
Table II. Comparison of demographics, timing of infection diagnosis, clinical presentation, management, and mortality between patients with previous thoracic open aortic repair and patients with previous thoracic endovascular aortic repair (TEVAR), among the 126 patients described in single-center case series or case reports SG infection, previous open repair (n ¼ 49)
EG infection, previous TEVAR, (n ¼ 77)
Availablea
Positiveb
%c
Male sex
30
25
Age, years, mean 6 SD
47
62.5 6 11.4
76
64.9 6 12.2
.17
Timing from index procedure, months, mean 6 SD
31
32.2 6 61.1
52
17.1 6 21.1
.03
83
Availablea
Positiveb
41
30
%c 73
P .39
Clinical presentation AEF
32
10
31
60
36
60
.01
ABF
32
4
13
60
11
18
.56
AEF þ ABF
32
0
0
60
1
2
Pain
37
27
73
48
17
35
<.001
Fever or chills
37
31
84
61
37
61
.02
Hematemesis
37
7
19
52
22
42
Hemoptysis
37
6
16
52
8
15
1.0
.02 1.0
Shock
37
4
11
52
8
15
Dyspnea
24
0
0
46
1
2
1.0
.75
Sepsis
24
2
8
55
6
11
1.0
Conservative
49
1
2
77
14
18.2
.02
Fistula repair only
49
1
2
77
10
12.9
.06
Surgical þ fistula repair
49
e
e
77
2
2.6
.52
Surgical fistula repair
49
2
4.5
77
7
9.1
.48
Endo þ fistula repair
49
12
27
77
23
29.9
.69
Endo fistula repair
49
25
52
77
16
20.8
.02
TEVAR / open
49
6
12.5
77
5
6.5
.35
30-day
48
16
33
73
19
26
1-year
47
27
57
67
36
54
.71
5-year
34
28
82
52
42
81
1.0
Managementd
Mortality .42
ABF, Aortobronchial fistula; AEF, aortoesophageal fistula; EG, endograft; SD, standard deviation; SG, surgical graft. Numbers in bold are considered statistically significant. a The number of patients for whom the specific information was available in the included studies. b The number of affected patients for each single item. c Percentages of positive patients of the available ones. d Conservative, medical treatment with or without minimally invasive additional procedures; Fistula repair only, esophageal or respiratory fistula repair, without associated aortic repair; Surgical þ fistula repair, surgical open aortic repair associated with additional maneuvers to repair the fistula; Surgical fistula repair, surgical open aortic repair without additional maneuvers to repair the fistula; Endo þ fistula repair, endovascular aortic repair (TEVAR) associated with additional maneuvers to repair the fistula; Endo fistula repair, endovascular aortic repair (TEVAR) without additional maneuvers to repair the fistula; TEVAR / open, endovascular aortic repair to obtain hemodynamic stability followed by open surgical aortic repair.
Staphylococcus epidermidis (1). Thirty-day mortality in patients with documented antimicrobial growth at culture was 14% (7/49), 5 of which were positive for MRSA. For MRSA-positive patients, 30-day mortality was 71%; for methicillin-sensitive S. aureus-positive patients, 9%; and for Streptococcus-positive patients, 8%. No significant difference in outcome was noted between patients with isolation of a single agent and patients with polymicrobial isolation. Conservative treatment, mainly represented by antibiotic therapy, sometimes associated with percutaneous
drainage of fluid collections or flushing, was performed in 2% of patients with SG (1/49) and 17% with EG (13/77). This approach resulted in a mortality rate of 100% at 30 days in the SG infection group and 38%, 75%, and 100% in the EG infection group at 30 days, 1 year, and 5 years, respectively. In the remaining patients, different surgical strategies were used, including aggressive open aortic repair with or without associated fistula repair, TEVAR with or without associated fistula repair, and esophageal or bronchial-pulmonary repair alone. Seventy-four patients
Journal of Vascular Surgery Volume
-,
Number
Kahlberg et al
7
-
Fig 2. Pooled odds ratio (OR) meta-analysis comparing 1-year mortality among patients with surgical graft (SG) infection and patients with endograft (EG) infection. CI, Confidence interval.
received in situ aortic repair, using a standard PET graft in 40 cases (antibiotic soaked in 15 cases), a cryopreserved allograft in 22 cases, a xenopericardial tube graft in 9 cases, a silver-coated PET graft in 2 cases, and an autologous vein graft in 1 case. Reported postoperative complications included respiratory failure in 22 patients, persistent sepsis in 13, multiorgan failure in 13, bleeding in 13, cardiac complications in 7, renal dysfunction in 5, paraplegia in 5, mediastinitis in 5, vocal cord paralysis in 3, esophageal leak in 2, wound infection in 2, recurrent aortopulmonary fistula in 1, encephalopathy in 1, and cholecystitis in 1. Global 30-day, 1-year, and 5-year mortality was 33%, 57%, and 82% in patients with SG infection and 26%, 54%, and 81% in patients with EG infection, respectively.
DISCUSSION Published scientific evidence on thoracic aortic prosthetic infection is definitely scarce. This is mainly due to the following reasons: d
d
d
d
Meta-analyses results. The OR meta-analysis of SG infection vs EG infection included five comparative studies (Fig 2). The statistical heterogeneity was low (I2 ¼ 2%; P ¼ .38). Even if it was not statistically significant, a slight tendency to increased overall mortality rate at 1 year was noted in patients with SG infection (pooled OR, 3.6; 95% confidence interval [CI], 0.9-14.7; P ¼ .073). The OR meta-analysis of conservative treatment vs surgical treatment included five comparative studies (Fig 3). The statistical heterogeneity was low (I2 ¼ 5%; P ¼ .37). No significant difference in overall mortality rate at 1 year was noted between conservative and surgical treatment (pooled OR, 0.6; 95% CI, 0.1-2.3; P ¼ .425). Finally, the OR meta-analysis of infected graft explantation vs preservation included seven comparative studies (Fig 4). The statistical heterogeneity was low (I2 ¼ 0%; P ¼ .42). A slight trend toward lower mortality rate at 1-year outcome was noted in patients with infected graft explantation (pooled OR, 0.3; 95% CI, 0.1-1.0; P ¼ .056).
Relative rarity of this complication. The majority of published papers are case reports or small series. This is probably also a cause of underestimation because of the lack of single-center reports by small centers experiencing few cases and not reporting them. Frequent late occurrence. Several patients might receive the diagnosis and treatment of infection years later in a different center from the index procedure. Heterogeneity in clinical presentation. Reports are often mixing elective and emergent cases, with or without associated AEF or ABF fistula. Heterogeneity in treatment strategies. Published papers usually include a number of different pharmacologic, surgical, and endovascular therapeutic options. Surgical strategies and timing modalities are mostly chosen according to the surgeon’s experience and the patient’s condition.
Infection of a thoracic aortic prosthesis is, however, a morbid condition, and mortality often exceeds 50% and can be as high as 100% for untreated cases.5-8 There are no prospective randomized trials concerning the treatment options for this condition, and the presence of many different variations makes it even more difficult to draw sound conclusions that may apply to the individual case. In this paper, we considered for analysis the original studies that referred to cases with an SG or EG involvement and not the reviews as a whole. However, a brief discussion of these reviews is provided here. In the last two decades, TEVAR has been gradually replacing open intervention with surgical replacement
8
Kahlberg et al
Journal of Vascular Surgery ---
2018
Fig 3. Pooled odds ratio (OR) meta-analysis comparing 1-year mortality among patients treated conservatively and patients treated surgically (any kind of surgical treatment). CI, Confidence interval.
Fig 4. Pooled odds ratio (OR) meta-analysis comparing 1-year mortality among patients with infected graft explantation and graft preservation. CI, Confidence interval.
of the aorta with a synthetic graft. Currently, in spite of the lack of prospective randomized trials, TEVAR is accepted as the treatment of choice for most cases of thoracic aortic disease that have anatomic feasibility; exceptions might apply mainly to patients with connective tissue diseases. Therefore, a relevant portion of the most recent data available on thoracic aortic graft infection actually refers to EG. Moulakakis et al48 in 2014 reviewed the literature for graft infection after TEVAR. They analyzed 55 patients treated with preservation of the EG and 41 with excision. Preservation of the EG with antibiotic therapy alone was shown not to be a durable option; their meta-analysis showed a trend for better outcomes in patients with EG excision. The authors found a trend for worse outcomes in patients with fistulas. Moreover, there was no difference whether the patients were treated with in situ or extra-anatomic reconstructions.
Our review identified 43 articles reporting on 49 patients with thoracic aortic SG infection and 184 patients with thoracic EG infection. Among these, four were multicenter studies, all focused on EG infection, associated with a fistula in 91% of cases. Patients presented with hematemesis or hemoptysis in 53% to 92% of cases and shock in 22% to 26%. Management included several different strategies that we may divide into “purely conservative” (19%-42% of cases) and somewhat “operative” (58%-81%). The operative cases included open aortic replacement, with or without associated esophageal or bronchial repair; repeated TEVAR, with or without associated esophageal or bronchial repair; and esophageal or bronchial open repair or endoscopic treatment (stenting) alone. Outcome reporting was totally heterogeneous in these multicenter reports, mainly represented by survival at different time points, and not specifying results per single management
Journal of Vascular Surgery Volume
-,
Number
Kahlberg et al
9
-
strategy in most cases. Nevertheless, conservative treatment was associated with the poorest survival (0%-20% at 1 year), whereas operative treatment showed a survival ranging from 27% to 39% at different follow-up points (up to 5 years). Among patients described in the 39 single-center studies, the time interval between index procedure and diagnosis of infection was shorter in patients with previous endografting than in patients with previous surgical repair. This result may be explained by the typically different cohort of patients treated with TEVAR compared with surgically treated ones. In particular, TEVAR is usually the preferred strategy for treatment of aortic pseudoaneurysms and saccular aneurysms, especially in an emergency, and a proportion of these cases may represent “undiagnosed” infected aneurysms or pseudoaneurysms. Moreover, the enhanced medical surveillance that is usually advised after TEVAR may account for a significant reduction in the time needed to diagnose infectious complications. Patients with EG infection were also more frequently reported to have an associated AEF than patients with SG infection. Even if detailed causes of this process are still unclear, this finding is consistent with a number of studies confirming that both thoracic and abdominal aortic endografting may be complicated by gastrointestinal fistulization, at least at the same rate as after surgical treatment.5,7,49 This review of single-center published series and case reports confirmed that clinical outcomes of thoracic aortic prosthetic infections are poor, with overall survival of 45% at 1 year and 19% at 5 years. Probably because of the small number of cases, meta-analyses failed to reach statistical significance in showing an undisputable difference in 1-year mortality between EG infection and SG infection. However, a slight (nonsignificant) trend was noted toward worse outcomes in case of SG infection that may be explained by the higher surgical complexity of “redo” interventions in case of previous open aortic surgery compared with patients who experienced only endovascular interventions, with presence of more extensive adhesions and “frozen” chest. Similarly, in comparing infected graft explantation vs graft preservation, the graft explantation strategy seemed to be associated with a better outcome, even if this result may be biased by the small number of studies included and by the low statistical power. Nonetheless, this finding is consistent with the previous published evidence on superiority of EG explantation vs preservation in patients with post-TEVAR infections,48 probably reflecting the inability of antibiotic therapy alone to eradicate infection involving either a surgical or an endovascular aortic prosthetic graft. We acknowledge some limitations of this review. The heterogeneity of the included patients (infection of surgical or endovascular graft, presence or absence of
associated fistula, conservative or surgical management) makes it difficult to draw consistent conclusions from pooling to drive clinical decision-making. However, consideration should be given to the results presented because they are at present the most comprehensive data on a topic that is still not sufficiently addressed in the literature. The decision to include a large number of small studies, including case reports, is driven by the rarity of this disease. This may account for publication bias due to the plausible attitude of small centers to report most “interesting” and “successful” cases. Also, a time bias should be considered in comparing results of SG infection with the more recent cases of EG infection. Furthermore, a number of studies were excluded for lacking or incomplete reporting.
CONCLUSIONS This systematic review showed that thoracic aortic EG infection presents some differences compared with SG infection regarding frequency of associated aortoesophageal fistulization, time interval between index procedure and diagnosis of infection, and symptoms at presentation. Overall survival is poor in both groups, but meta-analysis found a trend of worse outcome in patients with SG infection. Even if treatment strategies are various and depend on each center’s experience and on specific patients’ conditions, the current evidence probably seems to support surgical treatment with graft explantation when it is feasible. However, these data are so far too weak to justify changes in clinical practice, and more comparative studies are needed to provide clinical recommendations in the future.
AUTHOR CONTRIBUTIONS Conception and design: AK, GM, RC Analysis and interpretation: AK, AG, FV, NM, NC, GM Data collection: AK, AG, DL, GM Writing the article: AK, AG, DL, GM Critical revision of the article: AK, FV, NM, NC, GM, RC Final approval of the article: AK, AG, DL, FV, NM, NC, GM, RC Statistical analysis: AK, AG Obtained funding: Not applicable Overall responsibility: GM
REFERENCES 1. Kahlberg A, Melissano G, Mascia D, Loschi D, Grandi A, Chiesa R. How to best treat infectious complications of open and endovascular thoracic aortic repairs. Semin Vasc Surg 2017;30:95-102. 2. Moher D, Liberati A, Tetzlaff J, Altman DG; PRISMA Group. Preferred reporting items for systematic reviews and metaanalyses: the PRISMA statement. BMJ 2009;339:b2535. 3. Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 2000;283:2008-12.
10
Kahlberg et al
Journal of Vascular Surgery ---
4. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003;327:557-60. 5. Chiesa R, Melissano G, Marone EM, Marrocco-Trischitta MM, Kahlberg A. Aorto-oesophageal and aortobronchial fistulae following thoracic endovascular aortic repair: a national survey. Eur J Vasc Endovasc Surg 2010;39:273-9. 6. Smeds MR, Duncan AA, Harlander-Locke MP, Lawrence PF, Lyden S, Fatima J, et al. Treatment and outcomes of aortic endograft infection. J Vasc Surg 2016;63:332-40. 7. Czerny M, Eggebrecht H, Sodeck G, Weigang E, Livi U, Verzini F, et al. New insights regarding the incidence, presentation and treatment options of aorto-oesophageal fistulation after thoracic endovascular aortic repair: the European Registry of Endovascular Aortic Repair Complications. Eur J Cardiothorac Surg 2014;45:452-7. 8. Czerny M, Reser D, Eggebrecht H, Janata K, Sodeck G, Etz C, et al. Aorto-bronchial and aorto-pulmonary fistulation after thoracic endovascular aortic repair: an analysis from the European Registry of Endovascular Aortic Repair Complications. Eur J Cardiothorac Surg 2015;48:252-7. 9. Kieffer E, Sabatier J, Plissonnier D, Knosalla C. Prosthetic graft infection after descending thoracic/thoracoabdominal aortic aneurysmectomy: management with in situ arterial allografts. J Vasc Surg 2001;33:671-8. 10. Kieffer E, Chiche L, Gomes D. Aortoesophageal fistulae: value of in situ aortic allograft replacement. Ann Surg 2003;238: 283-90. 11. Eggebrecht H, Mehta RH, Dechene A, Tsagakis K, Kühl H, Huptas S, et al. Aortoesophageal fistula after thoracic aortic stent-graft placement. A rare but catastrophic complication of a novel emerging technique. JACC Cardiovasc Interv 2009;2:570-6. 12. Czerny M, von Allmen R, Opfermann P, Sodeck G, Dick F, Stellmes A, et al. Self-made pericardial tube graft: a new surgical concept for treatment of graft infections after thoracic and abdominal aortic procedures. Ann Thorac Surg 2011;92:1657-62. 13. Lemaire SA, Green SY, Kim JH, Sameri A, Parenti JL, Lin PH, et al. Thoracic or thoracoabdominal approaches to endovascular device removal and open aortic repair. Ann Thorac Surg 2012;93:726-33. 14. Lyons OT, Patel AS, Saha P, Clough RE, Price N, Taylor PR. A 14-year experience with aortic endograft infection: management and results. Eur J Vasc Endovasc Surg 2013;46: 306-13. 15. Canaud L, Alric P, Gandet T, Ozdemir BA, Albat B, MartyAne C. Open surgical secondary procedures after thoracic endovascular aortic repair. Eur J Vasc Endovasc Surg 2013;46: 667-74. 16. Kawamoto S, Sato M, Motoyoshi N, Kumagai K, Adachi O, Saito T, et al. Outcomes of a staged surgical treatment strategy for aortoesophageal fistula. Gen Thorac Cardiovasc Surg 2015;63:147-52. 17. Luehr M, Etz CD, Nozdrzykowski M, Garbade J, Lehmkuhl L, Schmidt A, et al. Emergency open surgery for aortooesophageal and aorto-bronchial fistulae after thoracic endovascular aortic repair: a single-centre experience. Eur J Cardiothorac Surg 2014;47:374-83. 18. Suzuki T, Kawamoto S, Motoyoshi N, Akiyama M, Kumagai K, Adachi O, et al. Contemporary outcome of the surgical management of prosthetic graft infection after a thoracic aortic replacement: is there a room to consider vacuumassisted wound closure as an alternative? Gen Thorac Cardiovasc Surg 2014;63:86-92. 19. Kahlberg A, Melissano G, Tshomba Y, Leopardi M, Chiesa R. Strategies to treat thoracic aortitis and infected aortic grafts. J Cardiovasc Surg (Torino) 2015;56:269-80.
2018
20. Coselli JS, Koksoy C, LeMaire SA. Management of thoracic aortic graft infections. Ann Thorac Surg 1999;67:1990-8. 21. Fatima J, Duncan AA, De Grandis E, Oderich GS, Kalra M, Gloviczki P, et al. Treatment strategies and outcomes in patients with infected aortic endografts. J Vasc Surg 2013;58: 371-9. 22. Kubota H, Endo H, Noma M, Ishii H, Tsuchiya H, Yoshimoto A, et al. Xenopericardial roll graft replacement for infectious pseudoaneurysms and graft infections of the aorta. J Cardiothorac Surg 2015;5:1-9. 23. VonFricken K, Karamanoukian HL, Ricci M, Taheri A, Bergsland J, Salerno TA. Aortobronchial fistula after endovascular stent graft repair of the thoracic aorta. Ann Thorac Surg 2000;70:1407-9. 24. Bond SE, McGuinness CL, Reidy JF, Taylor PR. Repair of secondary aortoesophageal fistula by endoluminal stentgrafting. J Endovasc Ther 2001;8:597-601. 25. Hance KA, Hsu J, Eskew T, Hermreck AS. Secondary aortoesophageal fistula after endoluminal exclusion because of thoracic aortic transection. J Vasc Surg 2003;37:886-8. 26. Porcu P, Chavanon O, Sessa C, Thony F, Aubert A, Blin D. Esophageal fistula after endovascular treatment in a type B aortic dissection of the descending thoracic aorta. J Vasc Surg 2005;41:708-11. 27. Abdul-Ghani A, Pisipati S, McWilliams R, Page RD. Aortobronchial fistula following aortic and bronchial stenting of a thoracic aneurysm. Eur J Cardiothorac Surg 2006;29:419-21. 28. Riesenman PJ, Farber MA, Mauro MA, Selzman CH, Feins RH. Aortoesophageal fistula after thoracic endovascular aortic repair and transthoracic embolization. J Vasc Surg 2007;46: 789-91. 29. Martens K, De Mey J, Everaert H, Delvaux G, van den Brande P. Aortoesophageal fistula following endovascular exclusion of a thoracic aneurysm. Int Angiol 2007;26: 292-6. 30. Civilini E, Bertoglio L, Melissano G, Chiesa R. Aortic and esophageal endografting for secondary aortoenteric fistula. Eur J Vasc Endovasc Surg 2008;36:297-9. 31. Isasti G, Gómez-Doblas JJ, Olalla E. Aortoesophageal fistula: an uncommon complication after stent-graft repair of an aortic thoracic aneurysm. Interact Cardiovasc Thorac Surg 2009;9:683-4. 32. Riesenman PJ, Farber MA. Management of a thoracic endograft infection through an ascending to descending extra-anatomic aortic bypass and endograft explantation. J Vasc Surg 2010;51:207-9. 33. Yavuz S, Kanko M, Ciftci E, Parlar H, Agirbas H, Berki T. Aortoesophageal fistula secondary to thoracic endovascular aortic repair of a descending aortic aneurysm rupture. Heart Surg Forum 2011;14:E249-51. 34. Lee SH, Song PS, Kim WS, Park KB, Choi S. A case of stent graft infection coupled with aorto-esophageal fistula following thoracic endovascular aortic repair in a complex patient. Korean Circ J 2012;42:366-8. 35. Yamashita S, Dohi T, Shimizu Y, Momomura S. Aortopulmonary fistula caused by an infected thoracic aortic false aneurysm rupturing after endovascular stent placement. Drug Discov Ther 2012;6:278-82. 36. Okamoto H, Tamenishi A, Matsumura Y, Niimi T. Composite vein graft reconstruction for infected descending aortic prosthesis. Ann Thorac Surg 2012;93:2061-3. 37. Akkoyunlu Y, Aslan T, Ceylan B, Inan B, Aydin C, Arvas G. Aortic stent infection in etiology of fever of unknown origin in a geriatric patient. Asian Pac J Trop Dis 2012;2(Suppl 2): S963-4. 38. Onodera M, Inoue Y, Fujino Y, Kikuchi S, Endo S. A case of secondary aortoesophageal fistula inserted a covered
Journal of Vascular Surgery Volume
39.
40.
41.
42.
43.
44.
-,
Number
Kahlberg et al
11
-
self-expanding esophageal stent to control gastrointestinal bleeding. Case Rep Gastrointest Med 2013;2013:857135. Muradi A, Yamaguchi M, Kitagawa A, Nomura Y, Okada T, Okita Y, et al. Secondary aortoesophageal fistula after thoracic endovascular aortic repair for a huge aneurysm. Diagn Interv Radiol 2013;19:81-4. M, Bioc ina B, Uzun S, Mestrovic T, Safradin Petrunic I. Cryopreserved aortic homograft for in situ replacement of infected thoracic stent graft associated with distal aortic arch rupture and hematemesis. Ann Thorac Surg 2014;98: 2219-21. Yamasaki M, Abe K, Misumi H, Ito J, Nakanishi Y. Endograft infection after hybrid surgery for chronic Stanford type B aortic dissection: endograft infection and treatment. Surg Today 2015;45:1575-8. Georvasili VK, Bali C, Peroulis M, Kouvelos G. Management of an aorto-esophageal fistula, complicating a descending thoracic aortic aneurysm endovascularly repaired. Gen Thorac Cardiovasc Surg 2016;64:216-9. Sueda T, Takahashi S, Katayama K. Successful treatment of an infected thoracic endovascular stent graft. Gen Thorac Cardiovasc Surg 2016;64:273-6. Afifi RO, Mushtaq HH, Sandhu HK, Khalil K, Safi HJ, Estrera AL. Successful multistaged surgical management of secondary aortoesophageal fistula with graft infection. Ann Thorac Surg 2016;101:e203-5.
45. Yamamoto M, Nishimori H, Iguchi M, Orihashi K. Pathological investigation of graft-related oesophageal fistula. Interact Cardiovasc Thorac Surg 2017;24:813-4. 46. Clarke NS, Murthy R, Reznik SI, Jessen ME. Coxiella burnettiassociated thoracic endovascular stent graft infection. J Card Surg 2017;32:506-7. 47. Sladojevic M, Bjelovic M, Ilic N, Mutavdzic P, Koncar I, Dragas M, et al. Open surgical treatment of secondary aortoesophageal and aortobronchial fistula after thoracic endovascular aortic repair and esophagocoloplasty in a second procedure. Ann Vasc Surg 2017;44:417.e11-6. 48. Moulakakis KG, Mylonas SN, Antonopoulos CN, Kakisis JD, Sfyroeras GS, Mantas G, et al. Comparison of treatment strategies for thoracic endograft infection. J Vasc Surg 2014;60:1061-71. 49. Kahlberg A, Rinaldi E, Piffaretti G, Speziale F, Trimarchi S, Bonardelli S, et al; MAEFISTO collaborators. Results from the Multicenter Study on Aortoenteric Fistulization After Stent Grafting of the Abdominal Aorta (MAEFISTO). J Vasc Surg 2016;64:313-20.e1. Submitted May 1, 2018; accepted Oct 26, 2018.
Additional material for this article may be found online at www.jvascsurg.org.
11.e1
Journal of Vascular Surgery
Kahlberg et al
---
2018
Supplementary Table (online only). Quality assessment of included studies with the Newcastle-Ottawa Scale Authors
Year
Selection (of 4)
Comparability (of 2)
Outcome (of 3)
Total No. of stars (of 9)
Study quality
VonFricken
2000
3
1
3
7
High
Bond
2001
3
1
3
7
High
Hance
2003
3
2
3
8
High
Porcu
2005
2
1
3
6
Moderate
Abdul-Ghani
2006
2
2
3
7
High
Riesenman
2007
2
2
3
7
High
Martens
2007
3
1
3
7
High
Civilini
2008
3
1
3
7
High
Isasti
2009
3
1
3
7
High
Riesenman
2010
2
1
3
6
Moderate
Yavuz
2011
3
1
3
7
High
Lee
2012
3
1
3
7
High
Yamashita
2012
3
1
3
7
High
Okamoto
2012
3
1
3
7
High
Akkoyunlu
2012
3
2
3
8
High
Onodera
2013
3
1
3
7
High
Muradi
2013
3
2
3
8
High
Petrunic
2014
3
1
3
7
High
Yamasaki
2015
3
1
3
7
High
Georvasili
2016
3
1
3
7
High
Sueda
2016
3
1
3
7
High
Afifi
2016
3
1
3
7
High
Yamamoto
2017
3
1
3
7
High
Clarke
2017
3
1
1
5
Moderate
Sladojevic
2017
3
1
3
7
High
Coselli
1999
4
2
3
9
High
Fatima
2013
4
2
3
9
High
Kubota
2015
4
2
3
9
High
Kieffer
2001
4
2
3
9
High
Kieffer
2003
4
2
3
9
High
Eggebrecht
2009
4
2
3
9
High
Czerny
2011
4
2
1
7
High
LeMaire
2012
4
2
3
9
High
Lyons
2013
4
2
3
9
High
Canaud
2013
4
2
3
9
High
Kawamoto
2014
4
2
3
9
High
Luehr
2014
4
2
3
9
High
Suzuki
2014
4
2
3
9
High
Kahlberg
2015
4
2
3
9
High
Chiesa
2009
4
2
3
9
High
Czerny
2014
4
2
3
9
High
Czerny
2015
4
2
3
9
High
Smeds
2016
4
2
3
9
High