Characteristics and Outcome of Vascular Graft Infections: A Risk Factor and Survival Analysis

Abstract Background Vascular graft infection (VGI) is a serious complication after implantation of arterial vascular grafts. Optimal surgical and pathogen-specific antimicrobial treatment regimens for VGI are largely unknown. We evaluated patients with arterial VGI according to onset, location, microbiological and imaging characteristics, and surgical and antimicrobial treatment and performed an outcome evaluation. Methods Consecutive patients with VGI treated in 2 hospitals from 2010 through 2020 were retrospectively analyzed. Uniform definition criteria and standardized outcome evaluation were applied. Logistic regression was used for multiple analysis; survival analysis was performed with Kaplan-Meier analysis and a log-rank test. Results Seventy-eight patients with VGI were included: 30 early-onset cases (<8 weeks after graft implantation) and 48 late-onset cases, involving 49 aortic and 29 peripheral grafts. The median time from initial implantation to diagnosis of VGI was significantly longer in aortic than peripheral VGIs (363 vs 56 days, P = .018). Late-onset VGI (odds ratio [OR], 7.3; P = .005) and the presence of surgical site infection/complication (OR, 8.21; P = .006) were independent risk factors for treatment failure. Surgical site infection/complication was associated with a higher risk for early-onset VGI (OR, 3.13; P = .040). Longer infection-free survival was observed in cases where the infected graft was surgically removed (P = .037). Conclusions This study underlines the importance of timely diagnosis of VGI and preventing surgical site infections/complications at graft implantation. It highlights the complexity of infection eradication, especially for late-onset infections, and the importance of adequate antimicrobial and surgical treatment.

The management of VGI requires complex surgical revisions and prolonged antimicrobial treatment.Treatment decisions typically depend on expert opinion or individual clinical experiences.Uniform recommendations regarding the surgical and antimicrobial management of VGI are lacking, and the optimal treatment concepts are unknown.It is unclear when VGIs can be treated successfully with graft removal, considering risk of extensive surgical intervention potentially associated with high perioperative morbidity and mortality [8,12,13].Thus, it is important to identify patient subpopulations and factors associated with beneficial or unfavorable impact on the treatment outcome of VGI.
In this study, we analyzed the characteristics of arterial VGI and identified risk factors for treatment failure of individual subgroups, including the impact of graft location, onset of infection, and type of antimicrobial and surgical treatment.
Organization ICD-10 codes Z95.88 (presence of other cardiac devices implants and grafts) and T82.7 (infection and inflammatory reaction due to other cardiac and vascular devices, implants, and grafts) were screened for study inclusion.Patients were included if the definition criteria for arterial VGI were fulfilled and the vascular graft was implanted at 1 of the 2 study centers.To keep the heterogeneity within the cohort at a minimum, we chose to exclude infections of endovascular grafts.Efforts are ongoing to analyze the endovascular graft infection cohort separately.Coronary bypass-associated infections were also excluded.This study was reviewed and approved by the institutional ethics committees of both study centers.

Infection Definition
The definition criteria applied to arterial VGI in this study were established before the publication of the MAGIC criteria and are summarized in Supplementary Table 1 [8,14].This decision was made to avert further heterogeneity within the cohort by applying nonuniform definition criteria.VGI was confirmed if 1 definite criterion or 2 suggestive criteria were present at the time of diagnosis.Low-virulent pathogens such as coagulasenegative staphylococci, Cutibacterium spp, Bacillus spp, or Corynebacterium spp were considered pathogens if detected in at least 2 independent microbiological samples.Aortic infection included thoracic, abdominal, and thoracoabdominal VGIs, while peripheral VGIs included grafts of the peripheral arteries.VGIs were divided into early onset (diagnosed <8 weeks after initial implantation) and late onset (at ≥8 weeks after implantation).Extracted information included patientspecific data (demographics, comorbidities, underlying vascular disease), infection-specific data (diagnostic criteria, microbiology), surgical and antimicrobial treatment, and follow-up evaluation regarding treatment outcome.

Outcome Evaluation
The follow-up period was defined as the period from the initial diagnosis of VGI to the last reported patient contact.An active follow-up with surviving patients was not pursued in this study.
The outcome was defined as treatment success if none of the following events occurred during the follow-up period: (1) reoperation at the vascular graft site; (2) infection-related death; or (3) persisting, recurring, or newly diagnosed VGI.Treatment failure was defined as the occurrence of any of these events.The infection-free survival interval was determined as the time from VGI diagnosis to the last reported patient contact at which a patient did not present with 1 of these events.

Evaluation of Adequacy of Antimicrobial Therapy
Initial empiric antimicrobial treatment consisted of a broadspectrum intravenous β-lactam antibiotic (eg, aminopenicillin with β-lactamase inhibitor or a cephalosporin of first, second, or fourth generation) in combination with a glycopeptide (eg, vancomycin, teicoplanin) or lipopeptide (daptomycin) antibiotic.In case of penicillin allergy, carbapenem (imipenem or meropenem) or fosfomycin was used to replace the penicillin derivative or cephalosporin.
Adequacy of treatment was evaluated according to antimicrobial adjustment to susceptibility testing (whether all pathogens were susceptible to at least 1 of the administered agents) and whether biofilm-active agents were used for sufficient treatment duration.Biofilm-active treatment was defined as administration of a rifampin-combination regimen for staphylococci, ciprofloxacin for gram-negative bacteria, ampicillin or amoxicillin for streptococci and enterococci, and metronidazole or clindamycin for anaerobic bacteria [6,[15][16][17][18].If no biofilm-active antibiotics were available for drug-resistant pathogen strains, antimicrobial suppression for >1 year to lifelong was considered adequate treatment [19,20].The adequacy of antimicrobial treatment was evaluated by 2 independent infectious disease physicians who were blinded to the treatment outcome.

Statistical Analysis
A sample size of 65 patients was estimated necessary for statistical analysis.A dropout rate of 10% to 15% was projected due to insufficient data, and a statistical power of 80% was estimated for 55 patients.SPSS (version 25.0; IBM) and RStudio (version 2023.06.1+524) were used for statistical analysis and plotting.For bivariate analysis, a chi-square test and Fisher exact test were applied for categorial variables and a Mann-Whitney U test for scaled variables.P < .050was considered statistically significant.Multiple analysis was performed as logistic regression.Variables were selected forward according to statistical and clinical significance.We limited the number of variables for multiple analysis by applying the rule of 10 [21].Survival analysis was performed with Kaplan-Meier analysis.A log-rank test was used to compare survival rates of different patient subcohorts.Perioperative infectious complications during the index implantation of the later-infected graft were significantly more often observed in aortic VGIs than peripheral VGIs (P = .019),while surgical site infection or complication was observed at similar rates in both subgroups (P = .585).There was no significant correlation between the presence of surgical site infection/ complication or perioperative infectious complication and the onset of VGI in the aortic subgroup (P = .129and P = .535)or the peripheral subgroup (P = .069and P = .060).

Infection Characteristics
Table 2 summarizes the infection characteristics of the analyzed cohort.Thirty patients (38.5%) presented with early-onset VGI and 48 (61.5%) with late-onset VGI.The median time from initial implantation to diagnosis of VGI was 160 days and was significantly longer in aortic VGIs than peripheral VGIs (363 vs 56 days, P = .018).Among all patients, the most frequent clinical symptoms of VGI were discharge (n = 22) or pain (n = 22) at the surgical site and fever (n = 19).Aortic VGI more often manifested with systemic signs of infection such as fever (P < .001),whereas graft thrombosis or stenosis (P = .003)was more frequently observed in peripheral VGIs.

Imaging Findings
Findings from radiologic and nuclear medical imaging are summarized in Table 2. Across all patients, visible perigraft fluid (n = 22), air (n = 15), and abscess (n = 15) surrounding the graft were the most frequently observed signs of VGI in diagnostic imaging.Supplementary Table 2 summarizes the rate of radiologic imaging suggestive of VGI.Computed tomography (CT) angiography and CT were the most frequently applied imaging methods (n = 50) and showed signs of VGI in 86.8%.If performed in aortic VGI, positron emission tomography/CT was positive for VGI in all cases, while all 3 cases of peripheral VGI that underwent such scans showed no enhancement (P = .038).

Microbiological Data
Table 3 summarizes the cohort's microbiological characteristics of VGI analyzed in this study.Coagulase-negative staphylococci and Staphylococcus aureus were the most frequently isolated pathogens, followed by enterobacteria and enterococci.Candida spp were identified in 17 cases (21.8%): 12 aortic VGIs and 5 peripheral VGIs (P = .426).Strikingly, while the prevalence of polymicrobial VGIs was 41.6% across the total cohort, all cases but 1 (Candida) were polymicrobial (P < .001),matching the theory of microbiome translocation caused by surgery or trauma [22].An increase of Candida-positive VGIs has been reported [19].However, the  influence of Candida as causative pathogens on the morbidity and mortality of patients with VGI remains unclear [23,24].
In our cohort, we found no correlation between the presence of Candida and the location of the infected graft (P = .426)or the reported outcome (P = .132).Of 14 culture-negative infections, 1 was located peripherally, and 13 were aortic VGIs (P = .009).
Multidrug-resistant pathogens were found in <10% of patients.Patients with polymicrobial infections received inadequate antimicrobial treatment significantly more often than those with monomicrobial infections (54.1% and 30.0%,P = .032).
Supplementary Table 3 summarizes the type of microbiological sample in which the pathogen was isolated.Intraoperative samples and explanted graft material grew pathogens significantly more often in peripheral VGIs than aortic VGIs (P = .005and P = .036,respectively).

Antimicrobial Treatment
In our cohort, all patients received antimicrobial treatment.The median time of initial antimicrobial therapy was 50 days (range, 12-189).A biofilm-active antibiotic was included in 28 cases (41.2%), and suppressive antimicrobial treatment was administered in 27 cases (36.5%).Appropriate biofilmactive agents and suppressive antimicrobial treatment were found in antimicrobial adequately treated patients significantly more often (both P < .001)but did not affect the outcome (P = .486and P = .844,respectively).New pathogens during treatment occurred in 18 cases (25.7%) and significantly more often in patients with inadequate antimicrobial therapy (39.4% and 13.5%, P = .013).Overall adequate antimicrobial treatment was observed in only 51.3% of the cohort.Inadequately treated infections consisted of 13 monomicrobial, 20 polymicrobial, and 4 culture-negative infections.The proportions of detected pathogens are similar to those observed in the total cohort, with staphylococci being the most frequently detected group.Of cases with inadequate antimicrobial treatment (n = 38), 20 patients received only 1 antimicrobial agent in the initial antimicrobial regimen.
For the primary end point of infection-free survival, 50% of all patients with adequate antimicrobial treatment later showed treatment success, as opposed to only 39.5% (P = .350) of patients with inadequate antimicrobial therapy.

Surgical Treatment
Figure 1 shows the study cohort according to surgical treatment mode and outcome.In the presented cohort, 37 patients underwent graft removal (34 anatomic and 3 extra-anatomic revascularizations); 34 patients received graft-retaining surgery with debridement of the infected field; and 7 patients did not receive any surgical intervention for VGI.Among surgically treated patients, 56.3% experienced peri-or postoperative complications within the first 10 days of surgery: significant bleeding (n = 13), peripheral ischemia (n = 8), and graft thrombosis (n = 6).Infectious complications occurred in 14 cases (19.7%), of which Postoperatively, patients with inadequate antimicrobial therapy showed significantly more infectious complications (P = .010),more often required escalation of surgical treatment (P = .008)and amputation (P = .002),and had a higher total number of surgical procedures (P = .013).

Outcome
The median time from VGI diagnosis to last patient contact or death was 6.87 months (range, 0.13-55.89).Thirty-five patients (44.9%) were infection-free at the time of last contact.Overall, only 44.9% of the cohort (n = 35) had successful eradication of VGI.This number was even lower in the subgroup of late-onset infections (35.4%, P = .034).
During follow-up, 18 deaths (23.1%) were reported, including 12 VGI-related deaths.Although this was a nonsignificant trend, the rate of VGI-related deaths was twice as high in patients with inadequate antimicrobial therapy (21.1% and 10.0%, P = .176).
Figure 2 visualizes the Kaplan-Meier analysis of infectionfree survival according to location of the infected graft, VGI onset, surgical mode, and adequacy of antimicrobial treatment.Onset, graft status, and adequacy of antimicrobial treatment did not have an impact on the infection-free survival of patients in our cohort.Only graft removal was associated with longer infection-free survival (Figure 2C; median [95% CI], 10.68 months [0-25.75]vs 26.61 months [4.34-48.89];P = .037).
To identify independent risk factors for treatment failure, we further performed a logistic regression.In this model applied to the whole study cohort, we identified surgical site infection or complication after the initial graft implantation (odds ratio, 8.2; 95% CI, 1.86-36.36;P = .006)and late-onset VGI (odds ratio, 7.3; 95% CI, 1.83-28.92;P = .005)as independent risk factors for treatment failure.

DISCUSSION
In this study, we present 78 retrospectively analyzed cases of arterial VGI according to anatomic location, onset, treatment, and outcome of VGI.Several of our findings are relevant to the clinical management of VGI and may affect the treatment outcome: (1) only about half of all analyzed patients received adequate antimicrobial therapy; (2) patients with removal of the infected graft had a longer infection-free survival; (3) aortic VGIs more often presented as late-onset infections than peripheral VGIs; and (4) late onset of VGI and surgical site infection or complication at the initial graft implantation are independent risk factors for treatment failure in VGI.
Only about half of the analyzed cohort received overall adequate antimicrobial therapy.Although this number is similar to previously reported cohorts, it is lower than expected [25].Possible reasons were the use of single-agent antimicrobial regiments, limited spectrum therapy in polymicrobial infections,  Management of Vascular Graft Infection • OFID • 5 and the lack of antifungal agents in empirical antimicrobial therapy.Due to increasing numbers of fungal VGI, the inclusion of an antifungal agent in the empirical treatment, especially for intracavitary VGI, is recommended [19,26].Patients with inadequate antimicrobial therapy showed higher rates of new pathogen growth during treatment and persisting infection and reinfection during follow-up, subsequently resulting in treatment failure in most cases.This subgroup also more frequently required surgical escalation, which in this highly comorbid cohort poses a higher risk for complications itself.To prevent the growth of new or untargeted pathogens, patients should receive broad initial empirical antimicrobial treatment, which can be de-escalated rather than escalated after microbiological testing.Like most published retrospective studies addressing VGI treatment, we are not able to provide a data-based recommendation for the exact duration of antimicrobial treatment.To overcome this shortcoming, prospective studies need to be conducted evaluating the exact duration of antimicrobial therapy for VGI.
In Kaplan-Meier analysis, patients who had their infected grafts removed showed significantly longer infection-free survival as compared with patients who did not have their grafts removed, highlighting the impact of surgical treatment decision making in the management of VGI.Similar results have been reported presenting favoring outcomes for patients with graft removal [10][11][12].A possible limitation to this result is the proportion of patients in the graft retention cohort not consenting to or being too unfit for extensive surgery, including 7 cases that did not receive any surgical therapy for VGI.Even though we found no significant differences in medical comorbidities or age between patients with and without graft removal in our cohort, factors not documented in this study, such as nonconsent to surgery, general frailty, altered mental state, or level of care, could imply a higher rate of patients likely to have bad outcomes without regard to the surgical regimen in the graft retention cohort.In these cases, previous studies have presented good results from less-invasive graft-retaining surgery, also in combination with prolonged antimicrobial therapy [8,25,27,28].Based on these results, a less-invasive graft-retaining surgical approach can be individually discussed with patients of elevated surgical risk, inoperability, or wish for less invasive surgery.However, our results strongly suggest that if operability is granted, removal of the infected graft should be the standard surgical approach in VGI.
In our study, patients with peripheral VGIs had an earlier onset of infection than patients with aortic VGIs.This finding is concordant with results from a previously published study [29].While a cause could be the more subtle clinical appearance of aortic VGIs possibly prolonging the time to diagnosis, further research on the reasons for a later onset of aortic VGI needs to be conducted [30].Our results show good radiologic and nuclear imaging accuracy for the detection of aortic VGIs.To improve the time to diagnosis of aortic VGI, we therefore suggest radiologic and nuclear medical imaging alongside microbiological culture as recommended by the MAGIC consortium to rule out or confirm VGI in patients with an inlaying aortic vascular graft presenting with a fever of unclear focus [31,32].
Surgical site infection has been previously determined as an independent risk factor for the development of VGI [1,2,4].To our knowledge, our study is the first to also determine surgical site infection or complication as a risk factor for treatment failure of VGI.Thus, reducing the number of surgical site infections and other surgical site complications at the initial implantation of vascular grafts is important to improve the management of VGI.
Another independent risk factor for treatment failure was a late onset of VGI, which in our study was surprisingly not influenced by surgical or antimicrobial treatment status.Thus, other influences must be considered.In nonsignificant trends, patients with late-onset VGI presented with a higher median age at diagnosis, twice as many VGI-related deaths, and 3 times as many reinfections as in early-onset VGIs.Age >70 years at diagnosis has been described as a risk factor for in-hospital mortality of patients with aortic VGIs [5,7].Another possible factor for treatment failure in the late-onset cohort is the insufficient eradication of mature biofilm.Mature biofilm is less susceptible to common antimicrobial agents and shows increased production when exposed to high shear forces and fluid flow rates, as can be found in the arterial vascular system [33][34][35][36][37][38][39].In our study, neither the inclusion of a biofilm-active agent nor overall adequate antimicrobial therapy had a significant impact on the outcome.Since scientific evidence for biofilm activity of antimicrobial agents other than rifampin and fluroquinolones is controversial, other approaches of biofilm-active treatment should be considered [17,18].Multimodal and novel approaches for biofilm eradication, including the use of bacteriophages, matrix-degrading enzymes, and improved graft material, should be considered in the treatment, especially for late-onset VGIs [37,[40][41][42][43][44].Besides sufficient biofilm eradication, further research needs to be conducted focusing on other causes for poor outcome of late-onset VGI.
In our study, reinfections or persisting infections during follow-up occurred in 31 cases, majorly contributing to the relatively large proportion of negative outcomes.To address this shortcoming, more studies are needed to establish validated follow-up regimes after VGI, such as further evaluation of the impact of nuclear imaging as a diagnostic parameter for therapy and relapse monitoring [45][46][47].
Our study is limited by its retrospective design, dependency on documented data in the hospital's electronic patient record, the inclusion of 2 hospitals with possible varying internal practices, the limited number of patients, and the limited median duration of follow-up.However, we applied uniform definition criteria for VGI, and both institutions followed the same treatment recommendations and were consulted by the same infectious disease consultation service.Even though the absolute number of patients in our study is small and limits detailed statistical analysis, we present one of the largest cohorts of patients with arterial VGI.We were able to assess the impact of location, onset, and treatment of infection, as well as new risk factors for treatment failure of VGI, and we present new findings that will ideally help to improve the challenging management of VGI in the future.Our results also stress the need for prospective study designs to further optimize the diagnostic workup and surgical and antimicrobial treatment of VGI.

Table 1 . Demographic Characteristics of the Study's Patient Cohort
Data are presented as No. (%) unless otherwise indicated.