¹⁸F-FDG PET/CT in left ventricular assist device infections: in-depth characterization and clinical implications

Abstract

Background:

Previous retrospective studies suggest a good diagnostic performance of ¹⁸F-FDG-PET/CT in left ventricular assist device (LVAD) infections. Our aim was to prospectively evaluate the role of PET/CT in the characterization and impact on clinical management of LVAD infections.

Methods:

40 patients (58[53–62] years) with suspected LVAD infection and 5 controls (69[64–71] years) underwent ¹⁸F-FDG-PET/CT. Four LVAD components were evaluated: exit site and subcutaneous driveline (peripheral), pump pocket, and outflow graft. The location with maximal uptake was considered the presumed site of infection. Infection was confirmed by positive culture (exit site or blood) and/or surgical findings.

Results:

Visual uptake was present in 40 patients (100%) in the infection group vs 4 (80%) control subjects. For each individual component, presence of uptake was more frequent in the infection than in the control group. The location of maximal uptake was most frequently the pump pocket (48%) in the infection group and the peripheral components (75%) in the control group. SUVmax was higher in the infection than in the control group: SUVmax (average all components):6.9[5.1–8.5] vs 3.8[3.7–4.3], p=0.002; SUVmax (location of maximal uptake):10.6±4.0 vs 5.4±1.9, p=0.01.

Pump pocket infections were more frequent in patients with bacteremia than without bacteremia (79% vs 31%, p=0.011). Pseudomonas (32%) and methicillin-susceptible Staphylococcus aureus (29%) were the most frequent pathogens and were associated with pump pocket infections, while Staphylococcus epidermis (11%) was associated with peripheral infections. PET/CT affected the clinical management of 83% of patients with infection, resulting in surgical debridement (8%), pump exchange (13%), and upgrade in the transplant listing status (10%) leading to 8% of urgent transplants.

Conclusions:

¹⁸F-FDG-PET/CT enables the diagnosis and characterization of the extent of LVAD infections, which can significantly affect the clinical management of these patients.

Introduction

Left ventricular assist devices (LVAD) have an expanding role in patients with congestive heart failure, and with their increasing use, there is a growing concern about infections^1,2. The International Society for Heart and Lung Transplantation(ISHLT) terms LVAD-specific infections as those related to the device hardware, including pump and/or cannula, pocket, and percutaneous driveline infections¹.

Imaging techniques can help in the diagnosis of LVAD infections; a transesophageal echocardiogram is recommended in patients with suspected infection, especially in those with bacteremia or fungemia. Computed tomography(CT) can help in the identification of infection in extracardiac components; however, infections can only be identified by CT when they are in more advanced stages and accompanied by other macroscopic changes, such as fluid collections, soft tissue stranding, or gas pockets^3,4. Moreover, the presence of metal objects, such as the LVAD pump, can lead to severe streaking and beam hardening artifacts that limit the reliability of the CT interpretation⁵. Molecular imaging techniques have gained interest in the diagnosis of cardiovascular infections, particularly in cases in which other imaging techniques are inconclusive⁶. One of the key benefits of ¹⁸F-Fluorodeoxyglucose Positron Emission Tomography(¹⁸F-FDG-PET) is in the detection of inflammatory cells early in the infection process, before morphological damage appears⁶. The initial changes in inflammation and infection include hyperemia with cellular migration, release of cytokines and increase in glycolysis that is the basis of ¹⁸F-FDG uptake, that may appear even before specific symptoms of the disease^7,8. Previous retrospective studies and meta-analysis have suggested a good diagnostic performance of ¹⁸F-FDG-PET/CT in LVAD infections^5,9–15. PET/CT allows extracardiac evaluation and can help with the characterization of the infection by identifying the involved components^5,14, which can have prognostic implications^10,14. Several causative pathogens can be involved in LVAD infections¹⁶, and PET/CT could be useful to understand the affinity of the different microorganisms for the distinct LVAD components. To date, the lack of a standardized PET/CT procedure and interpretation criteria has limited the confirmation of the utility of this technique.

In this prospective study, we aimed to evaluate the usefulness of ¹⁸F-FDG-PET/CT in the diagnosis and characterization of LVAD infections, its role in the identification of patterns associated with different pathogens involved, and its implications in clinical management.

Methods

Study population

LVAD recipients evaluated at Mount Sinai Hospital between December 2019 and January 2023 were offered participation in the study. Inclusion criteria were: age>20 years, known or suspected device infection (pain and/or purulent drainage in the driveline exit site, along the driveline tract, over the left chest wall, or fever with no other suspected focus of infection). Exclusion criteria included fasting glucose>250 mg/dl before imaging. Patients with clinical deterioration due to suspected worsening infection or reinfection were referred for a follow-up scan. The control group consisted of LVAD recipients that were evaluated in their routine visits and had no clinical suspicion of device infection (no pain or purulent drainage, no fever, no previous positive cultures).

The study was approved by the Institutional Review Board(HS-11–00216), and all patients provided written consent.

PET/CT acquisition

¹⁸F-FDG-PET was performed in a Siemens Biograph PET/CT system (Siemens Healthineers, Erlangen, Germany) in accordance with the guidelines of the Society for Nuclear Medicine and Molecular Imaging¹⁷. Dietary preparation included a high-fat, low-carbohydrate diet the day before the scan and an overnight fast before the scan. A standard ¹⁸F-FDG dose of 5MBq/kg was administered 60 minutes before imaging^17,18. Low-dose CT without intravenous contrast was obtained for attenuation correction and anatomic localization. The imaging protocol included 2–3 bed positions, with acquisition for 5 minutes/bed spanning the chest and abdomen to cover all LVAD components.

Low-dose non-contrast CT images were not evaluated independently from PET images due to their limited resolution. When clinically indicated, diagnostic CTs were assessed separately for evaluation of diagnosis of LVAD infections.

PET/CT analysis

Both attenuation and non-attenuation-corrected images were evaluated. Each scan was systematically assessed using Osirix-MD software (Osirix-imaging, Geneva, Switzerland) by 2 experts (AD,MGT) blinded to the clinical data for increased ¹⁸F-FDG uptake in four of the LVAD components: exit site, subcutaneous driveline, pump pocket, and outflow graft(Figure-1). Exit site and subcutaneous driveline were defined as peripheral components, and pump pocket and outflow graft as central components⁵. The presence of visual uptake was defined if an elevated signal was detected compared to surrounding tissue, and it was confirmed in non-attenuation-corrected images to avoid overestimation caused by metal¹⁹. For each LVAD component, multiple regions of interest were drawn to include areas with visual uptake. Conversely, in the absence of visual uptake, all of the LVAD components were measured, carefully excluding spillover from neighboring structures. We measured maximum standard uptake values (SUVmax) and tissue-to-background ratio (TBRmax), and corrected SUVmax for mean activity in the liver¹⁴. SUVmax(total) was calculated as the average SUVmax of the LVAD components per patient. Furthermore, SUVmax of the location with the highest uptake was reported (SUVmax(site)). The location of maximal uptake was considered the presumed site of infection. Presence of elevated uptake in thoracic lymph nodes or extracardiac foci suggestive of infection was also evaluated. Abnormal CT findings included soft tissue stranding, fluid, and/or air collection^3,5.

Figure 1. ¹⁸F-FDG uptake in the different LVAD components.

Representative examples of PET/CT images showing ¹⁸F-FDG uptake in the different LVAD components, in axial views (A-C) and coronal view (D). A) shows a patient with intense uptake in the exit site; B) shows uptake in the subcutaneous driveline; C) shows intense uptake in the pump pocket; and D) shows significant uptake in the outflow graft. Red arrows indicate the location of the uptake.

Clinical Data

Clinical data included demographics, transplant listing (UNOS-United Network for Organ Sharing) status and type of LVAD, blood and wound cultures, antimicrobial therapy and hospital admission. Patients were followed for changes in clinical management and events.

LVAD infection was confirmed by the presence of a positive wound culture or confirmed bacteremia with no other apparent focus of infection, or the presence of histologic features associated with infection or purulent drainage at surgery (surgical debridement or at the time of LVAD explantation), in accordance with ISHLT criteria¹.

Statistical analysis

Continuous variables were described as mean ±standard deviation when normally distributed, otherwise median(Q1-Q3). Categorical variables were presented as n(%). Comparisons between groups were assessed using the Fisher exact test for categorical variables, the independent samples t-test and the Wilcoxon rank-sum test for continuous variables with normal or non-normal distributions, as appropriate. A p-value<0.05 was considered statistically significant. Diagnostic performance and optimal thresholds for SUVmax and TBRmax were established using the receiver operation characteristic curve, the area under the curve(AUC), and the Youden Index. All analyses were performed using Stata version 16.1(StataCorp, College Station, Texas).

Results

40 patients with suspected LVAD infection (58[53–62]years, 32.5% female) and 5 controls (69[64–71]years, 20.0% female) were included. Baseline characteristics are summarized in Table-1. The mean glucose level at the time of the ¹⁸F-FDG injection was 111.1±32.1mg/dl. LVAD infection was confirmed in all patients in the infection group.

Table 1.

Baseline characteristics in the infection group versus the control group.

	Infection	Control	p-value
n	40	5
Age, years, median [Q1-Q3]	58.0 [53.0 – 62.0]	69.0 [64.0 – 71.0]	0.09
Female, n (%)	13 (32.5)	1 (20.0)	1
Ethnicity, n (%)
African American	15 (37.5)	3 (60.0)	0.4
Caucasian	9 (22.5)	1 (20.0)	1
Asian	3 (7.5)	0 (0.0)	1
Hispanic	13 (32.5)	1 (20.0)	1
Smoking, n (%)	26 (65.0)	1 (20.0)	0.1
Body mass index, kg/m2, mean ± SD	29.7 ± 7.0	28.7 ± 8.3	0.8
Diabetes, n (%)	27 (67.5)	2 (40.0)	0.2
Etiology of cardiomyopathy, n (%)
Non-ischemic cardiomyopathy	26 (65.0)	5 (100)	0.3
Ischemic cardiomyopathy	13 (32.5)	0 (0.0)	0.3
Mixed cause	1 (2.5)	0 (0.0)	1
Transplant listing (UNOS status), n (%)
Not listed	19 (47.5)	1 (20.0)	0.4
UNOS status 3	1 (2.5)	0 (0.0)	1
UNOS status 4	20 (50.0)	4 (80.0)	0.4
LVAD indication, n (%)
Destination therapy	24 (60.0)	1 (20.0)	0.2
Bridge to transplant	16 (40.0)	4 (80.0)	0.2
Time since LVAD implantation, years, median [Q1 - Q3]	1.7 [1.1 – 3.0]	2.1 [1.0 – 3.8]	1
LVAD type, n (%)
Heart Mate 2	5 (12.5)	0 (0.0)	1
Heart Mate 3	28 (70.0)	5 (100)	0.3
Heart Ware	7 (17.5)	0 (0.0)	0.6
Blood parameters
White blood cell count, 103 cells/μL, median [Q1 – Q3]	7.1 [5.2 – 10.1]	5.6 [3.3 – 6.2]	0.07
C-reactive protein, mg/L, median [Q1 – Q3]†	39.0 [32.0 – 82.1]	N/A	N/A
Erythrocyte segmentation rate, mm/h, mean ± SD††	62.1 ± 48.8	N/A	N/A
Positive culture, n (%)	38 (95.0)	0 (0.0)	N/A
- Positive wound culture, n (%)	33 (82.5)	N/A	N/A
- Confirmed bacteremia, n (%)	14 (35.0)	N/A	N/A
Infection requiring admission, n (%)	30 (75.0)	N/A	N/A
Intravenous antibiotic prior to PET, n (%)	25 (41.0)	N/A	N/A

^*IQR= interquartile range; LVAD= left ventricular assist device; UNOS=United Network for Organ Sharing.

^†Available in 9 patients.

^††Available in 7 patients.

Infection versus control group

¹⁸F-FDG visual uptake was present in all patients in the infection group (100%,n=40;Table-2): uptake was present in the exit site in 37 patients (92.5%); in the subcutaneous driveline in 36(90.0%); pump pocket in 35(87.5%); and outflow graft in 29(72.5%). 32 patients (80.0%) presented hypermetabolic thoracic lymph nodes and 16(40.0%) other extracardiac uptake suggestive of infection(Supplemental-Table-1). Among the 35 patients with pump pocket uptake, 33(94.3%) also had uptake in the subcutaneous driveline. In the control group, visual uptake was present in 4 patients (80.0%): 3(60.0%) at the exit site and the subcutaneous driveline, 1(20.0%) in the pump pocket, and 3(60.0%) in the outflow graft. The presence of visual uptake in the pump pocket was significantly more frequent in the infection than in the control group (87.5% vs 20.0%,p=0.004).

Table 2.

PET/CT imaging characteristics in the infection group versus the control group.

	Infection	Control	p-value
n	40	5
Presence of visual uptake, n (%)	40 (100)	4 (80.0)	0.11
Exit site, n (%)	37 (92.5)	3 (60.0)	0.09
Subcutaneous driveline, n (%)	36 (90.0)	3 (60.0)	0.1
Pump pocket, n (%)	35 (87.5)	1 (20.0)	0.004
Outflow graft, n (%)	29 (72.5)	3 (60.0)	0.6
SUVmax (total), median [Q1-Q3]	6.9 [5.1 – 8.5]	3.8 [3.7 – 4.3]	0.002
TBRmax (total), median [Q1-Q3 ]	2.9 [2.1 – 3.7]	1.8 [1.4 – 1.9]	0.0 1
SUVmax per components
SUVmax exit site, median [Q1-Q3]	6.6 [4.5 – 9.7]	3.5 [3.3 – 4.5]	0.03
SUVmax subcutaneous driveline, median [Q1-Q3]	5.4 [3.7 – 11.2]	2.8 [2.8 – 2.9]	0.03
SUVmax pump pocket, median [Q1-Q3]	7.4 [5.4 – 10.5]	4.6 [3.9–5.3]	0.02
SUVmax outflow graft, median [Q1-Q3]	4.3 [3.3 – 5.3]	3.6 [3.1 – 3.9]	0.13
Location of maximal uptake			0.7
Peripheral (exit site or subcutaneous driveline), n (%)	19 (47.5)	3 (75.0)
Pump pocket, n (%)	19 (47.5)	1 (25.0)
Outflow graft, n (%)	2 (5.0)	0 (0.0)
SUVmax(site), mean ± SD	10.6 ± 4.0	5.4 ± 1.9	0.007
TBRmax(site), mean ± SD	4.5 ± 1.7	2.4 ± 0.9	0.01

^*SUVmax= maximum standard unit value; TBRmax= maximum target to background ratio

SUVmax(total) was higher in the infection compared to the control group (6.9[5.1–8.5] vs 3.8[3.7–4.3],p=0.001), as was TBRmax(total)(2.9[2.1–3.7] vs 1.8[1.4–1.9],p=0.004)(Figure-2). When evaluated by LVAD components within the infection group, the pump pocket had the highest uptake (SUVmax:7.4[5.4–10.5]) followed by the peripheral components (exit site and subcutaneous driveline, SUVmax:6.6[4.5–9.7] and 5.4[3.7–11.2], respectively), while the outflow graft exhibited the lowest values (SUVmax:4.3[3.3–5.3]). SUVmax was higher in all components in the infection group compared to the control group, except for the outflow graft(Table-2).

Figure 2. ¹⁸F-FDG uptake in patients with infection versus control.

Representative PET/CT images of a patient with LVAD infection (left panel) and a control patient with no suspected LVAD infection (right panel). The left panel shows uptake in the exit site (blue arrow), subcutaneous driveline (white arrows) and pump pocket (red arrows). The right panel shows no uptake in any of the LVAD components. An implantable cardioverter-defibrillator generator and bicameral leads are observed in this patient, with no evidence of uptake. Median [Q1 – Q3] SUVmax(total) and TBRmax(total) are represented for the infection and control groups; the difference in these values between groups was statistically significant (p=0.002 for SUVmax(total); p=0.01 for TBRmax(total)).

In the infection group, 47.5% of the patients had the maximal uptake (presumed site of infection) in the peripheral components, and 52.5% in the central components (47.5% pump pocket; 5.0% outflow graft). On the contrary, in the control group, the maximal uptake was more frequently located in the peripheral components (n=3, 75.0%) than in the central components (n=1, 25.0%, pump pocket). SUVmax(site) was higher in the infection group compared to the control group (10.6±4.0 vs 5.4±1.9,p=0.007), a finding that was also observed for TBRmax(site)(4.5±1.7 vs 2.4±0.9,p=0.01). SUVmax(site) showed excellent performance in distinguishing patients with and without infection, with an AUC of 0.89; similar performance was observed for TBRmax(site) (AUC=0.85).

Bacteremia and causative pathogens

14 patients (35.0%) presented with bacteremia. The presumed site of infection differed between the groups with and without bacteremia: the most common location were the pump pocket (78.6%) in the group with bacteremia and the peripheral components (65.4%) in the group without bacteremia (p=0.011)(Table-3,Figure-3). There were no significant differences in SUVmax(site) and TBRmax(site) between patients with and without bacteremia.

Table 3.

PET/CT in patients with and without bacteremia

	Bacteremia	No Bacteremia	p-value
n (% of total n in infection group)	14 (35)	26 (65)
Location of maximal uptake, n (%)			0.011
Peripheral components, n (%)	2 (14.3)	17 (65.4)
Pump pocket, n (%)	11 (78.6)	8 (30.8)
Outflow graft, n (%)	1 (7.1)	1 (3.9)
SUVmax(site), mean ± SD	11.6 ± 4.7	10.1 ± 3.6	0.3
TBRmax(site), mean ± SD	4.7 ± 1.8	4.4 ± 1.7	0.5

^*SUVmax= maximum standard unit value; TBRmax= maximum target to background ratio

Figure 3. Location of ¹⁸F-FDG uptake according to the presence or absence of bacteremia.

Representative PET/CT images in a patient with bacteremia by methicillin-resistant Staphylococcus aureus (left panel, coronal view) and in a patient without bacteremia (right panel, axial view). Patient with bacteremia (left panel) presented intense uptake in the pump pocket (red arrows) while patient without bacteremia (right panel) presented intense uptake in the subcutaneous driveline (red arrows). Pump pocket is the most frequent presumed site of infection in patients with bacteremia, while it is the peripheral components in patients without bacteremia.

The pathogens most frequently found were Pseudomonas (31.6%) and methicillin-susceptible Staphylococcus aureus(MSSA)(29.0%). Other microorganisms involved are represented in Figure-4. High uptake was observed in both Pseudomonas (SUVmax(site):10.5±4.7) and MSSA(SUVmax(site):11.4±2.5), and both were frequently associated with pump pocket infections (50.0% of Pseudomonas and 45.5% of MSSA). Staphylococcus epidermidis infections (10.5%) also presented high uptake (SUVmax(site):11.3±5.6) but were more frequently associated with peripheral components (75.0%). All other infections caused by gram-positive microorganisms were located at the central components, while the remaining gram-negative infections were mostly associated with peripheral components(Table-4).

Figure 4. Microorganisms involved and distribution according to LVAD components.

Left panel shows the distribution of the different microorganisms involved. Right panel shows the percentage of microorganisms found in each LVAD component: peripheral components (exit site and subcutaneous driveline); pump pocket; and outflow graft. *MSSA= methicillin-susceptible Staphylococcus aureus; MRSA= methicillin-resistant Staphylococcus aureus; S.epidermidis= Staphylococcus epidermidis

Table 4.

PET/CT characteristics according to microorganisms involved.

	n (%)	SUVmax(site) mean ± SD	Peripheral components, n (%)	Pump pocket, n (%)	Outflow graft, n (%)
Pseudomonas	12 (31.6)	10.5 ± 4.7	5 (41.7)	6 (50.0)	1 (8.3)
MSSA	11 (29.0)	11.4 ± 2.5	6 (54.6)	5 (45.5)	0 (0.0)
Staphylococcus epidermidis	4 (10.5)	11.3 ± 5.6	3 (75.0)	1 (25.0)	0 (0.0)
Other gram-positive	5 (13.2)	9.5 ± 4.5	0 (0.0)	4 (80.0)	1 (20.0)
Other gram-negative	5 (13.2)	9.2 ± 2.2	4 (80.0)	1 (20.0)	0 (0.0)
Candida	1 (2.6)	8.4 ± 0.0	0 (0.0)	1 (100.0)	0 (0.0)

^*MSSA= Methicillin-susceptible Staphylococcus aureus; SUVmax= maximum standard unit value.

Clinical management and outcomes

Patients were followed-up for 1.2[0.3–1.8] years after PET/CT. 33 patients (82.5%) in the infection group had a clinical management change after PET/CT due to infection-related complications(Table-5). 21 patients (52.5%) had a change in antimicrobial therapy (15 were changed to intravenous therapy and 6 had a change in the antimicrobial), 3(7.5%) underwent surgical debridement and 5(12.5%) underwent pump exchange. Time from PET/CT to surgical intervention was 10[9–148] days. Moreover, 4 patients (10.0%) had an upgrade in the urgency status for transplant listing after PET/CT; 2 of them (5.0%) were transplanted after 3 and 32 days, respectively. All patients that underwent surgery or urgent transplant after PET/CT (n=10) presented exit site or subcutaneous driveline uptake, and 8 of them presented pump pocket uptake. The presence of driveline infection was confirmed in all of them at the time of surgery; the presence of pump pocket infection was confirmed in 7/8 patients with pump pocket uptake. To note, surgical intervention occurred 194 days after PET/CT in the patient with pump pocket uptake and no evidence of infection at surgery, and intravenous antimicrobial therapy was used during that time. In addition, patients that underwent surgery or urgent transplant after PET/CT had a trend to exhibit higher SUVmax and TBRmax than those who had medical management (SUVmax(total):8.9 vs 7.3,p=0.16; TBRmax(total):3.7 vs 3.1,p=0.19).

Table 5.

Clinical management after PET/CT.

	All infections	Peripheral infection	Central infection	p-value
n, (%)	40 (100)	19 (47.5)	21 (52.5)
Infection-related change in management during follow-up, n (%)	33 (82.5)	16 (40.0)	17 (42.5)	0.8
Antibiotic course altered, n (%)	21 (52.5)	10 (25.0)	11 (27.5)
Surgical exploration, n (%)	3 (7.5)	1 (2.5)	2 (5.0)
Pump exchange, n (%)	5 (12.5)	2 (5.0)	3 (7.5)
Upgrade in transplant listing, n (%)	4 (10.0)	3 (7.5)	1 (2.5)
Clinical outcomes
Hospital admission related to infection, n (%)	30 (75.0)	13 (32.5)	17 (42.5)	0.5
Heart transplant, n (%)	15 (37.5)	9 (22.5)	6 (15.0)	0.3
Death, n (%)	10 (25.0)	4 (10.0)	6 (15.0)	0.7

^*UNOS=United Network for Organ Sharing

During follow-up, 15 patients (37.5%) received a heart transplant in the infection group (time PET/CT to transplant: 0.9[0.5–1.7]years;Supplemental-Table-2), and 10 patients (25.0%) died after 0.6[0.3–1.4] years(Supplemental-Figure-1).There were no significant differences in the location of the infection in patients that received a heart transplant or died(Table-5). In the control group, one patient received a transplant (2.1 years after PET/CT), and there were no deaths during follow-up.

Follow-up scans

6 patients had a follow-up scan due to clinical suspicion of worsening of infection or reinfection 116[98–590] days after the first scan(Table-6). The presence of uptake in the different components overlapped between the first and second scan. In the first scan, visual uptake was present in the exit site in 6 patients (100%), in the subcutaneous driveline in 4(66.7%), in the pump pocket in 5(83.3%), and in the outflow graft in 5 patients (83.3%); the same prevalences were observed in the second scan except for the outflow graft (3 patients,50%). SUVmax(total) and TBRmax(total) increased from baseline to follow-up scans(Figure-5). The most frequent pathogen was Pseudomonas(50.0%); between the first and second scan, some patients had changes in antibiotic therapy (n=4) or surgical debridement of the exit site (n=2). One patient had a third PET/CT 220 days after the second because of clinical worsening. The distribution of visual uptake was the same as in the second scan but SUVmax(total) and TBRmax(total) were increased(Supplemental-Table-3). These findings led to higher status in the transplant list and eventually to cardiac transplantation, 52 days after the third PET/CT.

Table 6.

Comparison between first and follow-up PET/CT imaging.

	First PET/CT	Second PET/CT
n	6	6
Presence of visual uptake, n (%)	6 (100)	6 (100)
Exit site, n (%)	6 (100)	6 (100)
Subcutaneous driveline, n (%)	4 (66.7)	4 (66.7)
Pump pocket, n (%)	5 (83.3)	5 (83.3)
Outflow graft, n (%)	5 (83.3)	3 (50.0)
SUVmax(total), median [Q1-Q3]	5.2 [4.7 – 7.6]	5.7 [7.9 – 9.2]
TBRmax(total), median [Q1-Q3]	1.8 [1.7 – 3.8]	2.5 [2.1 – 2.9]

^*SUVmax= maximum standard unit value; TBRmax= maximum target to background ratio

Figure 5. Baseline and follow-up PET/CT scans in a patient with suspicion of reinfection.

Representative PET/CT images of a patient that underwent a baseline scan (left panel) and a follow-up scan (right panel) due to clinical worsening. In the baseline scan, there is uptake in the exit site (blue arrow), subcutaneous driveline (white arrows) and pump pocket (red arrows). Uptake is seen more intensely in the same LVAD components in the follow-up scan: exit site (blue arrow), subcutaneous driveline (white arrow) and pump pocket (red arrows). SUVmax(total) and TBRmax(total) for each scan are represented. A right-sided implantable cardioverter-defibrillator generator is observed in both scans, with no evidence of uptake.

Comparison with CT imaging

30 patients underwent a separate clinically indicated CT for assessment of infection, 23 of them (76.7%) were performed with contrast(Table-7). CT identified 10 of the LVAD infections (33.3% of the total performed CTs); 9 of them were in the peripheral components (versus 13 identified by PET/CT), and only one subject was found to have evidence of infection in the pump pocket, compared to 15 identified by PET/CT in this subgroup(Figure-6;Supplemental-Table-4).

Table 7.

PET/CT vs stand-alone CT for diagnosis of LVAD infections

	CT	PET/CT
n	30	30
Iodinated contrast, n (%)	23 (76.7)	0 (0.0)
Diagnosis of infection, n (%)	10 (33.3)	30 (100)
Location of infection, n (%)
Peripheral infection, n (%)	9 (90.0)	13 (43.3)
Pump pocket, n (%)	1 (10.0)	15 (50.0)
Outflow graft, n (%)	0 (0.0)	2 (6.7)

Figure 6. PET/CT versus CT for the evaluation of LVAD infections.

The left panel shows PET and PET/CT images of a patient with pump pocket infection. Intense uptake is observed involving the pump pocket (red arrows). Pacemaker leads are observed with no evidence of uptake. The right panel shows a contrast CT axial view of the same patient, at the level of the LVAD pump; no significant collection or abnormal finding is observed around the pump pocket. Note the streak artifact around the LVAD pump (red *), which limits the interpretation of CT images.

Discussion

The results of our prospective study show that LVAD infections can be accurately characterized by means of ¹⁸F-FDG-PET/CT, especially with the use of semi-quantitative techniques. PET/CT imaging also enables the identification of the different patterns of involvement within the LVAD components, according to the presence or absence of bacteremia and/or specific pathogens. Lastly, the information provided by PET/CT imaging can impact clinical management (i.e., changes in antimicrobials or urgent heart transplant listing).

Previous studies have shown that LVAD infections could be accurately detected by PET/CT^5,10,20. Consistent with these reports, the presence of visual uptake was identified in all patients with suspected LVAD infection. Nonetheless, the presence of visual uptake alone was not conclusive for the diagnosis of LVAD infection as it was found in some control patients as well; rather, our results suggest that PET/CT findings should be combined with semi-quantitative parameters in order to increase diagnostic accuracy. SUVmax and TBRmax were significantly higher in patients with infection compared to controls. Based on the location of the maximal uptake, the optimal threshold to distinguish patients with and without infection was established at 5.6 for SUVmax (92.5% (95%CI: 80.1%−97.4%) sensitivity and 80.0% (95%CI: 37.6%−96.4%) specificity) and 3.2 for TBRmax (70.0% (95%CI: 54.6%−81.9%) sensitivity and 80.0% (95%CI: 37.6%−96.4%) specificity)(Graphical-Abstract). Previously published retrospective studies identified similar thresholds^11,12. Optimal SUVmax thresholds for each component were established at 4.5 for exit site; 3.1 for subcutaneous driveline; 5.7 for pump pocket; and 4.3 for outflow graft (Supplemental-Table 5). The limited number of controls precludes taking these cut-off points as definitive for the general population, and the thresholds estimated in this study should be refined in larger studies. Visual uptake was common in the peripheral components in both infection and control patients, which might be linked to other factors besides infection, like, for example subcutaneous inflammation due to exit site manipulation. Previous studies have shown that ¹⁸F-FDG uptake can be found in sterile inflammation²¹; establishing the difference between infection and inflammation in the peripheral components can be achieved by means of semi-quantitative evaluation together with clinical suspicion (driveline exit purulent drainage, pain, etc.). Conversely, the presence of uptake in the pump pocket was almost exclusively found in the infection group, suggesting that PET/CT is a reliable tool for identifying pump pocket infections. This is contrary to what has been reported in previous retrospective studies, that had questioned the utility of PET/CT in the identification of pump pocket infections¹³. Despite the limitations of this technique related to metal artifacts¹⁹, we found significant differences between the infection and control groups regarding the frequency and intensity of pump pocket uptake. When compared to CT, the latter could only identify one case of pump infection out of 15 cases identified by PET/CT in this subgroup (6.7%); similarly, previous studies had shown that only 11% of all infections are identified by CT⁵. The follow-up scans also showed an excellent reproducibility of PET/CT in identifying the involved components. In addition, there is increased mortality when LVAD infections are complicated by bloodstream infections²². Our results suggest that pump pocket infections were frequently associated with the presence of bacteremia. On this basis, PET/CT could be considered as part of the diagnostic algorithm in patients with suspected LVAD infection and bacteremia to assess for the presence of pump pocket infection (Graphical-Abstract).

Graphical Abstract.

Patients with infection (upper left panel) had higher SUVmax and TBRmax than controls (upper right panel). Optimal thresholds for the differentiation between patients with infection and controls were identified at SUVmax(site) 5.6 and TBRmax(site) 3.2. In patients with suspected LVAD infection, the presence of bacteremia should raise the concern for pump pocket involvement (lower left panel). A PET/CT could help identify a pump pocket infection that may require a different therapeutic strategy. Frequently microorganisms involved in pump infections were Pseudomonas and methicillin-susceptible Staphylococcus aureus (MSSA). Patients with suspected LVAD infection but without bacteremia have more frequent involvement of the peripheral components of the LVAD (exit site and subcutaneous peripheral); frequent microorganisms found in this type of infections were Staphylococcus epidermidis.

Furthermore, different types of causative pathogens will result in variable severity of infection²³. Pump pocket infections were more frequent in patients with Pseudomonas and MSSA infections; the presence of these pathogens could be an additional marker for the selection of patients to undergo PET/CT imaging to rule out the presence of pump pocket infection. On the contrary, the presence of Staphylococcus epidermidis appears more likely to suggest an infection of the peripheral components (Graphical-Abstract). PET/CT affected the clinical management of patients diagnosed with LVAD infection. Information provided by PET/CT could facilitate surgical debridement, as well as pump exchange. The presence and extent of the infection identified by PET/CT were confirmed in patients undergoing surgery. Moreover, the results of PET/CT led to high urgency listing that facilitated heart transplant in three patients, which is similar to what has been previously described²⁴. Nonetheless, contrary to previous studies, we did not find any differences in outcomes between patients with peripheral and central infections⁵. This could be explained by the relatively small sample size and the lower rates of transplant and mortality in our cohort⁵.

Limitations

The presumed site of infection was defined as the location of maximal uptake, thus less intense uptake present in other LVAD components was deemed “less severe infection”, and might have confounded the original site of infection. The presence of infection in the inflow cannula might have generated false negative findings. Some of the patients in the infection group had received intravenous antimicrobial therapy before PET/CT. The study sample was relatively small, and the number of control patients limited; indeed, a larger control group might have resulted in greater differences. Lastly, compared to prior studies, ours overcame some of their limitations by virtue of being prospective and enrolling a broader and more heterogenous population.

In conclusion, ¹⁸F-FDG-PET/CT is a useful tool that allows the identification and characterization of the patterns and extension of LVAD infections according to the different causative pathogens and clinical scenarios. The use of this technique can help the clinician in the decision-making process for the management of these patients, identifying those who require surgical intervention, device exchange, or urgent heart transplants.