Author + information
- Received October 24, 2017
- Revision received January 11, 2018
- Accepted January 29, 2018
- Published online March 14, 2018.
- Jongho Kim, MD, PhDa,b,
- Erika D. Feller, MDc,
- Wengen Chen, MD, PhDa,
- Yuanyuan Liang, PhDd and
- Vasken Dilsizian, MDa,∗ ()
- aDepartment of Diagnostic Radiology and Nuclear Medicine, University of Maryland School of Medicine, Baltimore, Maryland
- bDivision of Nuclear Medicine and Molecular Imaging, Department of Radiology, Columbia University Medical Center, New York Presbyterian Hospital, New York, New York
- cDepartment of Medicine, Division of Cardiology, University of Maryland School of Medicine, Baltimore, Maryland
- dDepartment of Epidemiology and Public Health, University of Maryland School of Medicine, Baltimore, Maryland
- ↵∗Address for correspondence:
Dr. Vasken Dilsizian, Department of Diagnostic Radiology and Nuclear Medicine, University of Maryland School of Medicine, 22 South Greene Street, Baltimore, Maryland 21201.
Objectives The feasibility of 18F-fluorodeoxyglucose positron emission tomography/computed tomography (FDG PET/CT) for the diagnosis of left ventricular assist device (LVAD) infection has been demonstrated. Beyond the diagnoses of LVAD infection, the authors hypothesized that the pattern and site of the infection along its various components may significantly impact clinical management and patient outcome.
Background In patients with end-stage heart failure, the clinical use of LVAD for destination therapy is on the rise, accompanied by a higher prevalence of infections and serious complications.
Methods FDG PET/CT was performed in 35 heart failure patients with LVAD; 24 with and 11 without clinical suspicion of infection. Microbiology and/or clinical follow-up were used as the final diagnosis standard. Survival rates were compared in patients with and without FDG evidence of infection, and in relation to peripheral (exit wound site or driveline) versus central (cannula or pump) device infection.
Results Of 35 patients, 28 (80%) showed metabolic evidence of LVAD infection: 5 limited to the periphery and 23 with extension to the central components of the device. The remaining 7 patients showed no metabolic evidence of infection, which was confirmed by microbiology and clinical follow-up. When CT images were interpreted independently from the FDG PET and clinical information, only 4 of 35 (11%) suggested the possibility of infection. Fourteen of 28 (50%) infected patients died during a mean of 23 months of follow-up after the diagnosis by FDG PET/CT: 12 (86%) with central infection and only 2 with peripheral infection. By contrast, none of the 7 (0%) noninfected patients died (p = 0.03).
Conclusions FDG PET/CT is a useful technique for identifying LVAD infection and determining the site and pattern of the infection. The latter has clinical management and patient outcome implications.
One of the consequences of the constant shortage of donor hearts for cardiac transplantation is the expanded role of left ventricular assist device (LVAD) in the management of advanced heart failure both from bridge to transplantation and destination therapy (1). Although the implantation of the device is lifesaving, approximately 20% to 40% of the patients develop sepsis approximately 1 to 2 years after LVAD implantation (2–4). LVADs are powered externally through a percutaneous driveline (exit site at the abdominal wall), which could be a potential site for introducing bacteria into the bloodstream during daily activity. Driveline infections are the most frequent type of LVAD infections. Once the infection extends centrally, into the cannula and pump pocket, eradication of infection from the LVAD system is difficult. As such, early diagnosis of peripheral infection is desirable (5). Driveline infection can be effectively treated by reducing trauma by immobilization, antibiotics, and/or surgical debridement. On the other hand, infection that has extended to the central components requires explant of the LVAD and urgent heart transplantation.
The clinical diagnosis of LVAD infection with current diagnostic approaches remains challenging. Local infection signs (i.e., erythema or external suppuration) on physical examination cannot localize the infection source relative to the device (i.e., superficial or deep involving the central components of the device). Blood culture may not isolate the pathogen (particularly when antibiotics have been initiated), nor localize the infection site. Computed tomography (CT) findings are often nonspecific for infection, and metal device artifacts limit its sensitivity as well. Although radiolabeled white blood cell scintigraphy has been used for cardiac implantable electronic device infections (6), the images are relatively count poor, and its role for LVAD infections has not been established (7). Inconclusive diagnosis can lead to delayed antibiotic therapy or surgical debridement, which may have dire consequences to the patient.
Among patients with suspected cardiac mechanical device or prosthetic valve infection, recent publications advocate the use of 18F-fluorodeoxyglucose (FDG) positron emission tomography (PET) with CT (hybrid PET/CT). FDG is taken up by inflammatory cells (macrophages, neutrophils, and lymphocytes) at the site of infection and/or inflammation via glucose transporter (GLUT), such as GLUT1 and GLUT4, which are overexpressed in these cells (8). Given the high spatial and target-to-background contrast resolution of FDG PET/CT, recent publications advocate the application of FDG PET/CT for the detection of cardiac implantable device infections (9–12), as well as prosthetic valve endocarditis (13). A potential advantage of FDG PET/CT is in its detection of inflammatory cells early in the infectious process, before morphological damages follow (12,14).
Our laboratory recently demonstrated the feasibility of FDG PET/CT for the diagnosis of LVAD infection (15). In the current study, we determined whether FDG PET/CT imaging may: 1) localize the site of LVAD infection as peripheral (percutaneous exit wound site and/or driveline) versus central (cannula and/or pump pocket) infection; and 2) predict the clinical outcomes of patients with suspected LVAD infection. We hypothesized that patients who exhibit FDG PET/CT evidence of LVAD infection, particularly with central components of infection, would have worse clinical outcome than those without infection, or with infection limited to the peripheral components of the device.
Materials and Methods
The study group consisted of a total of 35 consecutive heart failure patients with LVADs (male/female = 28/7; 54 ± 11 years of age; ischemic/nonischemic cardiomyopathy = 11/24; LVAD type: HeartMate II [Thoratec, Pleasanton, California] in 15, HeartWare [HeartWare International, Framingham, Massachusetts] in 12, and Jarvik [Jarvik Heart, New York, New York] in 8) from nearly 82 total patients with LVAD, who underwent FDG PET/CT for evaluation of LVAD infection. Of the 35 patients studied, 24 (69%) were referred for suspected LVAD infection and the remaining 11 (31%) had baseline post-LVAD placement imaging without signs or symptoms of infection. The study was reviewed and approved by the University of Maryland Institutional Review Board.
FDG PET/CT acquisition and processing
All patients were provided a list of food and instructed to consume a low-carbohydrate/high-fat dinner. The patients fasted overnight before the scan, without breakfast. Fatty acid loading or intravenous administration of unfractionated heparin before image acquisition was not applied. Imaging was performed approximately 1 h after injection of 474 ± 63 MBq (12.8 ± 1.7 mCi) of FDG, in a Siemens Biograph PET/CT system (Siemens Healthineers, Erlangen, Germany). Low-dose CT without intravenous contrast was obtained for attenuation correction and anatomic localization. The imaging was limited to the chest and abdomen, covering the LVAD device and its components including the percutaneous exit wound site and driveline in the abdomen. Images were processed using iterative reconstruction. The mean serum glucose level at the time of the FDG injection was 106 ± 39 mg/dl.
Interpretation of FDG PET/CT images
Both attenuation- and nonattenuation-corrected images were interpreted by 2 nuclear medicine physicians (J.K. and W.C.) blinded to the clinical data. Discrepancy was resolved by a third nuclear medicine physician with consensus. Focal increased FDG uptake along any of the components of the device (exit wound site to the pump), with moderate to significantly higher intensity signal than the surrounding tissue that was confirmed on nonattenuation-corrected images was defined as infection. Linear homogeneous activity along the device, with intensity slightly higher than the surrounding tissue was defined as noninfectious or sterile inflammation. Infection in the exit wound site in the abdominal wall, and/or along the driveline in the abdominal subcutaneous region was defined as peripheral infection. Infection in the central components of LVAD (inflow/outflow cannula and/or pump pocket) is defined as central infection. Among patients who had concomitant cardiac implantable electronic device, such as pacemaker or automatic implantable cardioverter defibrillator, FDG uptake was also assessed at these sites. All patients were closely followed up in terms of treatment with antibiotics, surgical debridement, LVAD explant, emergent heart transplant due to LVAD infection, and death.
CT findings were also interpreted independently, blinded to the FDG PET results. Patients exhibiting abnormal CT findings, such as soft tissue stranding, fluid and/or air collection were classified as abnormal (compatible with infection), and those without abnormal CT findings were classified as normal. It is important to point out, however, that the presence of metal objects in the scan field (such as the LVAD pump) can lead to severe streaking and beam hardening artifacts that degrade the quality of the images and limit the reliability of the CT interpretation.
Patients’ characteristics were described using summary statistics, and the differences among the 3 groups (central infection, peripheral infection, and noninfection) were compared using Kruskal-Wallis H test for continuous variables and Fisher exact test for categorical variables. All-cause mortality rates were compared among the 3 groups using Fisher exact test. The Kaplan-Meier product-limit estimator was used to measure the fraction of patients living for a certain amount of time after the time of PET/CT scan. The log-rank test was used to compare Kaplan-Meier survival curves among the 3 infection groups. All statistical tests were conducted at a 2-sided significance level of 5%, and all analyses were performed using Stata/SE version 14 (StataCorp LP, College Station, Texas).
Data in all 35 consecutive heart failure patients with LVAD and FDG PET/CT scans regarding age, sex, ischemic versus nonischemic cardiomyopathy, LVAD type, indication for FDG PET/CT, time interval from LVAD implantation to imaging, interpretation of FDG PET findings, independent interpretation of CT findings alone (blinded to the PET results), follow-up interval to patient outcome in days, patient outcome, and microbiology data are listed in Table 1. FDG PET/CT localized infection in 28 of 35 patients (80%): 5 limited to periphery (exit wound site and/or driveline) and 23 with extension to the central components of the LVAD (cannula and/or pump pocket). The remaining 7 of 35 patients showed no FDG evidence of infection either at the LVAD or cardiac implantable electronic device pocket. Of note, 6 of 11 patients without clinical suspicion of infection were found to have FDG PET/CT evidence of infection: 3 with peripheral infections and 3 with central infections, but none were detected on the localizing CT (Table 1). Four of the 6 infected patients subsequently died, which included all 3 central infection cases. The most common pathogen isolated was Staphylococcus aureus.
The CT images were interpreted, blindly, independent from the FDG PET and clinical information. It is important to note that only 4 of 35 (11%) CT studies suggested the possibility of infection such as soft tissue stranding (#10), fluid or fluid with air (#14, #15, and #35); none in the control subjects. Interpretation of the CT portions showed significant streaking and beam hardening artifacts in all 35 cases, which limit evaluation of infection. The mean duration from LVAD implantation to FDG PET/CT imaging was 15 months (SD = 11) and the mean follow-up after the PET/CT was 12 months (SD = 13).
The central, peripheral, and noninfected groups did not differ in age (p = 0.35, Kruskal-Wallis H test), sex (p = 0.48, Fisher exact test), etiology of cardiomyopathy (p = 0.06, Fisher exact test), and LVAD type (p = 0.16, Fisher exact test), as shown in Table 2.
All-cause mortality rate differed significantly among infected and noninfected groups. None of the 7 (0%) noninfected patients died. By contrast, 14 of 28 (50%) infected patients died (p = 0.03, Fisher exact test). Among the 14 patients who died with LVAD infection, 12 (86%) had extensive infection involving the central components of the LVAD, involving the cannula exterior and or pump pocket, whereas only 2 had peripheral infection (p = 0.01, Binomial probability test). The survival distribution differed significantly among central, peripheral, and noninfected groups (p = 0.03, log-rank test), as shown in Figure 1. Fourteen of the 35 (40%) patients subsequently underwent cardiac transplantation, and none of the transplanted patients died.
Figure 2 shows representative cases of LVAD infection, with and without attenuation correction, in the peripheral percutaneous exit wound (Figure 2A), along the driveline (Figure 2B), central portion of the cannula exterior (Figure 2C), and the pump (Figure 2D).
The current study shows the utility of FDG PET/CT for detecting and localizing LVAD-related infections in heart failure patients, which represents the first and one of the largest series of its kind. Patient outcome differed significantly among central (cannula and pump pocket), peripheral (exit wound site or driveline), and noninfected groups. None of the noninfected patients died. By contrast, 50% of the infected patients died. Among the patients who died with LVAD infection, 86% had extensive infection involving the central components of the LVAD.
LVAD-related bacterial infections and sepsis are prevalent and often start at the superficial exit wound site (1–5). If not diagnosed and treated early, the exit site infection may progress over months to become deep tissue infections and move to the central components of the device. In clinical practice, reducing trauma by immobilization of the percutaneous driveline, treatment with antibiotics and surgical approaches (such as double tunnel) are effective in managing and treating driveline infections. Although long-term antibiotic treatment may be the necessary option in many patients, this often leads to drug-resistant organisms, especially in biofilm-producing bacteria, such as S. aureus and Pseudomonas (16). The pathogens from driveline culture in the current study were comparable with prior reports in driveline infections. Qualified LVAD-infected patients can successfully receive heart transplantation with survival outcomes that parallel to those of other transplant recipients (17).
Among an estimated 200,000 symptomatic end-stage heart failure patients on maximally tolerated medications who are candidates for heart transplantation, approximately 2,200 patients receive heart transplantation per year (18). Given the significant shortage of suitable donor hearts, the application of LVADs in these individuals provide improved functional capacity and quality of life, as well as meaningful improvement in survival, typically 3 to 5 years. However, patient outcome is limited by frequent adverse events of device infection, pump thrombosis, bleeding, and stroke (19,20). The most recent eighth annual Interagency Registry for Mechanically Assisted Circulatory Support report showed that among more than 20,000 patients, the overall survival remains >80% at 1 year and 70% at 2 years (21). Accordingly, it is important to develop an imaging biomarker that can identify LVAD infection early, when it is still limited to the peripheral components of the LVAD, for example, exit wound site or driveline. Misclassification or delay in the diagnosis of LVAD infection may require driveline debridement, LVAD explanation, or urgent transplantation, which negatively impact patient morbidity and mortality. As demonstrated in the current study, 86% of patients who died had extensive infection involving the central components of the LVAD. These findings highlight the importance of LVAD infection and its dire consequences if not detected and treated early, before extending to the central components of the LVAD. Thus, FDG PET/CT may guide medical and surgical management of LVAD-infected patients and predict patient outcome.
Prior publications have shown the utility of FDG PET/CT for detection of other cardiovascular implantable electronic device infections (e.g., pacemakers and prosthetic valves) and for guiding treatment (9,10,22). Similar to our experience in LVAD patients, it has been shown that it is not the mere presence or the intensity of metabolic activity, but rather the pattern of FDG uptake that more reliably distinguishes infection from noninfection in device pockets or synthetic vascular grafts (23,24). Infected electronic devices tend to have a heterogeneous pattern, with areas of more focal and intense FDG uptake, often accompanied by computed CT abnormalities on anatomic colocalized PET/CT imaging. By contrast, noninfected electronic devices tend to exhibit either photopenia (lack of metabolic signal) or a more homogeneous pattern of FDG uptake, which is diffuse, without focal areas of intense signals. Although cardiac CT alone may be helpful in some situations, it remains purely anatomic and often nonspecific for infection and affected by significant streaking and beam hardening artifacts from the metal device. In our study, when CT images were interpreted independently from the FDG PET and clinical information, only 11% suggested the possibility of infection.
Whereas a few patients had FDG PET/CT studies without suspicion of infection, the large majority of the patients in the study were referred for FDG PET/CT with a clinical concern for driveline exit infection and/or bacteremia with ongoing antibiotic coverage. As such, it is possible that referral bias may have contributed to the high mortality rate observed among the 35 LVAD patients. Although the number of LVAD patients studied and the number of the death events are relatively small, nonetheless, it represents the largest series of FDG PET/CT for LVAD infection evaluation in the published reports.
FDG PET/CT imaging differentiates LVAD infected from noninfected patients and further localizes the site of the infection as peripheral (percutaneous exit site and/or drive line) versus central (cannula exterior and/or pump pocket). Such early detection of LVAD infection has important clinical, therapeutic, and prognostic implications.
COMPETENCY IN MEDICAL KNOWLEDGE: LVAD-related bacterial infections and sepsis are prevalent and often start at the superficial exit wound site, and if not diagnosed and treated early, they may progress to become deep tissue infections and move to the central components of the device with dire prognostic consequences. FDG PET/CT imaging differentiates LVAD infected from noninfected patients and further localizes the site of the infection as peripheral (percutaneous exit site and/or driveline) versus central (cannula exterior and/or pump pocket).
TRANSLATIONAL OUTLOOK: Early detection and localization of LVAD infection with FDG PET/CT has important clinical, therapeutic, and prognostic implications. Although the number of LVAD patients studied and the number of the death events are relatively small, nonetheless, it represents the first and the largest series of FDG PET/CT for LVAD infection evaluation in the published reports.
The authors have reported that they have no relationships relevant to the contents of this paper to disclose. James Udelson, MD, served as the Guest Editor for this paper.
- Abbreviations and Acronyms
- computed tomography
- glucose transporter
- left ventricular assist device
- positron emission tomography
- Received October 24, 2017.
- Revision received January 11, 2018.
- Accepted January 29, 2018.
- 2018 American College of Cardiology Foundation
- Trachtenberg B.H.,
- Cordero-Reyes A.,
- Elias B.,
- Loebe M.
- Erba P.A.,
- Sollini M.,
- Conti U.,
- et al.
- Litzler P.Y.,
- Manrique A.,
- Etienne M.,
- et al.
- Jamar F.,
- Buscombe J.,
- Chiti A.,
- et al.
- Sarrazin J.-F.,
- Philippon F.,
- Tessier M.,
- et al.
- Dilsizian V.,
- Bacharach S.L.,
- Beanlands S.R.,
- et al.
- Dilsizian V.,
- Achenbach S.,
- Narula J.
- Gewirtz H.,
- Dilsizian V.
- Kim J.,
- Feller E.D.,
- Chen W.,
- Dilsizian V.
- Pinney S.P.,
- Anyanwu A.C.,
- Lala A.,
- et al.
- Gordon R.J.,
- Weinberg A.D.,
- Pagani F.D.,
- et al.,
- Ventricular Assist Device Infection Study Group
- Kirklin J.K.,
- Naftel D.C.,
- Kormos R.L.,
- et al.
- Dilsizian V.
- Dilsizian V.,
- Chandrashekhar Y.