Cardiac implantable electronic devices (CIED) figure in a broad clinical spectrum of infections, such as local CIED infections, which can appear as isolated local infections (LI) or associated with systemic lead infections (SI). SI involve endovascular lead and intracardiac lead infections, including infective endocarditis (IE). General diagnosis is challenging and is based on microbiological data and cardiac imaging techniques such as transesophageal echocardiography (TEE).1–3,518F-fluorodeoxyglucose-positron emission tomography/computed tomography ([18F]FDG-PET/CT) has improved the diagnostic evaluation of prosthetic valve endocarditis and has been incorporated as a major diagnostic criterion in guidelines.1 In addition, it has been recently shown that hypermetabolism of the spleen and bone marrow (BM) as detected by [18F]FDG-PET/CT can be considered an indirect sign of IE in native or prosthetic valves.4,6

Despite the latest evidence, the overall usefulness of [18F]FDG-PET/CT in CIED infections requires further characterization. Several cohort studies have been published,3,5 showing high diagnostic yield for generator pocket infections but much lower performance in lead-associated infection (SI).7 TEE is also unable to detect lead vegetations in many patients with bacteremia, who probably have an endovascular lead infection (SI).2 [18F]FDG-PET/CT could help to improve the diagnosis in all topographical regions of CIEDs, including in endovascular leads, which cannot be accessed by TEE.

The primary endpoint of this study was to determine the diagnostic yield of [18F]FDG-PET/CT in each of the different CIED topographical regions: pocket, subcutaneous, endovascular, and intracardiac leads. Secondary endpoints were to analyze the performance of [18F]FDG-PET/CT compared with that of TEE in diagnosing SI, to define the diagnostic value of spleen and BM hypermetabolism as an indirect sign of SI, and to assess the potential usefulness of [18F]FDG-PET/CT in the follow-up of CIED infections without complete device removal and suppressed with chronic antibiotics to avoid relapses, guiding physicians on when to stop chronic oral suppression (CAS) therapy.

We included 54 cases and 54 controls; the characteristics of the 2 groups are presented in table 1. In 25% of cases, less than 152 days elapsed between the implant or device change procedure and the clinical infection.

Comparison between cases with isolated local infection and those with systemic infection or both types of infection

Cases were divided into those with isolated LI (n=13) and those with SI with or without local infection (n=41). Baseline characteristics were similar between the w2 groups (table 2). Local signs of device infection were present in 87% (47/54) of cases: 100% (13/13) with isolated LI and in 82.9% (34/41) of those in the SI group (P <.01). Of the patients with SI, 34.1% (14/41) had a positive echocardiography result. Microbiological positivity and etiology were distributed homogeneously in the 2 groups, with a predominance of Staphylococcus aureus and coagulase-negative staphylococci (CNS) (table 1 of the supplementary data). The specific classification of SI in terms of the diagnostic criteria is summarized in table 2 of the supplementary data. Patients with SI underwent significantly more removal surgery (70.7% vs 38.4%; P=.04); those with isolated LI received more CAS (61.5% vs 24.4%; P <.01). There were no statistically significant differences between patients with isolated LI and SI regarding reimplant surgery, in-hospital mortality, or relapse. There were no differences in [18F]FDG-PET/CT results globally or for any topographical segment during the interval between CIED implant/replacement and [18F]FDG-PET/CT (< 3 months vs> 3 months). All characteristics comparing groups and [18F]FDG-PET/CT results are summarized in table 3 of the supplementary data.

[18F]FDG-PET/CT accuracy results

The main results can be found in table 3. The overall sensitivity of [18F]FDG-PET/CT for confirmed CIED infection was 85% (46/54). Pocket sensitivity was 79% (37/47), subcutaneous lead 57% (27/47), endovascular lead 22% (9/41), and 10% (4/41) intracardiac lead. However, intracardiac lead sensitivity might be underestimated because 31.5% (17/54) of cases showed unsuccessful myocardial inhibition. The negative predictive value was 15% (8/54). Median time on antibiotic treatment before [18F]FDG-PET/CT acquisition was 5 [0-14] days in cases with positive results and 13 [5–16] days in cases with negative results (P=.19). Despite the existence of a trend, there were no significant differences were found regarding the period between antibiotic was initiated and [18F]FDG-PET/CT performance; 12 (22.2%) cases had been on antibiotic therapy prior to [18F]FDG-PET/CT acquisition with a median duration of 6 [0.0–14.0] days.

Table 3.

Overall diagnostic accuracy of [18F]FDG-PET/CT according to the 4 topographical regions of CIED infection

	CIED infectionn=54	Pocket infectionN=47*	Subcutaneous leadN=47*	EndovascularleadN=41	Intracardiac leadN=41
Sensitivity	85%(75.5, 94.5)	79%(66.7, 90.7)	57%(43.0, 71.8)	22%(9.9, 34.9)	10%(0.5, 18.2)
Specificity	100%(93.4, 100.0)	100%(92.4, 100.0)	100%(92.4, 100.0)	100%(91.3, 100.0)	NA
Positive predictive value	100%(93.4, 100.0)	100%(92.4, 100.0)	100%(92.4, 100.0)	100%(91.3, 100.0)	100%(91.3, 100.0)
Negative predictive value	87%(77.9, 96.3)	84.4%(74.3, 94.5)	73%(60.6, 85.4)	62.8%(48.4, 77.2)	59.3%(45.7, 72.9)

CIED, cardiac implantable electronic device; NA, not available; LI, isolated local infection.

*

13 isolated LI cases + 34 SI with LI.

Figure 1 shows positive [18F]FDG uptake examples and sensitivity values of FDG-PET/CT in a visual 3-dimensional representation of each CIED topographical region.

Figure 1.

Central illustration. The figure shows examples of positive FDG uptake and sensitivity values of [18F]FDG-PET/CT in a 3D visual representation of each CIED topographical region: pocket (blue), subcutaneous (green), endovascular (yellow), and intravascular (red). 3D, three-dimensional; [18F]FDG-PET/CT, 18F-fluorodeoxyglucose-positron emission tomography/computed tomography. CIED, cardiac implantable electronic-device.

(0.68MB).

Table 4 compares diagnostic performance between TEE and [18F]FDG-PET/CT in patients with systemic infection showing fever, leukocytosis and elevated C-reactive protein with positive blood cultures or positive lead cultures/16SrRNA-PCR and/or positive echo. In those patients, when [18F]FDG-PET/CT was combined with TEE, the definite diagnosis rate of infection significantly increased from 34% (14/41) to 56% (23/41) (P=.04) due to detection of endovascular involvement, with rates higher in the bacteremic (from 38.8% ([7/18] to 66.7% [12/18]) than in the nonbacteremic form (from 30.4% ([7/23] to 47.8% [11/23]) of systemic infections (P=.37).

Table 4.

Diagnostic performance of [18F]FDG-PET/CT

Type of systemic infection	Transesophageal echocardiography
	Positive	Negative	Total
With bacteremia (N=18)
Endovascular [18F]FDG-PET/CT
Positive	2a	5	7 (38.9)
Negative	5a	6	11
Total	7 (38.8)	11	18
Intracardiac [18F]FDG-PET/CT
Positive	2	0	2 (11.1)
Negative	5	11	16
Total	7 (38.8)	11	18
Without bacteremia (N=23)
Endovascular [18F]FDG-PET/CT
Positive	0	2b	2 (8.7)
Negative	7a	14	21
Total	7 (30.4)	16	23
Intracardiac [18F]FDG-PET/CT
Positive	0	2b	2 (8.7)
Negative	7	14	21
Total	7 (30.4)	16	23

[18F]FDG-PET/CT, 18F-fluorodeoxyglucose-positron emission tomography/computed tomography.

Values are expressed as absolute numbers or No. (%).

Diagnostic performance of [18F]FDG-PET/CT compared to transesophageal echocardiography in 41 patients with systemic infection with (18 patients) or without (23 patients) bacteremia.

a

Patients simultaneously have vegetations on the leads and/or tricuspid valve.

b

These were different patients.

ROC curves were analyzed for the median SUVmax of all 4 CIED topographical regions and the ratio between each SUVmax/liver SUVmean and blood pool-SUVmean. Clinically significant values were only found in pocket uptake for SUVmax and SUVmax/SUVmean liver values, it is shown in figure 2. The remaining ROC curves can be found in Figures 1 to 5 of the supplementary data.

Figure 2.

A: ROC curve for CIED pocket SUVmax cutoff point 2.35 [sensitivity: 79.63%; 92.59%]. B: ROC curve for CIED pocket SUVmax/SUVmean liver, cutoff point 1.28 [sensitivity: 75.56%; specificity: 88.89%]. CIED, cardiac-implantable-electronic-device; ROC, SUV, standardized uptake value, SUVmax, maximum standardized uptake value.

(0.16MB).

Several cohort studies of CIED infections have been published in recent years,7,11,12 reporting high sensitivity and specificity values for [18F]FDG-PET/CT in pocket infections but lower diagnostic performance in lead-associated infections. However, to date there is no gold standard for assessing the subcutaneous and endovascular lead portion in CIED infections. In addition, differentiation between LI and SI may be problematic, as intraoperative lead contamination in patients with LI might occur during device extraction.2–4,11.

In our study, [18F]FDG-PET/CT demonstrated an overall sensitivity for CIED infections of 85%: 79% for pocket infections, and 57% for subcutaneous lead infections. In contrast, in line with previous studies,7,12 our results showed low sensitivity on endovascular (22%) and intracardiac leads (10%). The specificity of [18F]FDG-PET/CT was 100% for all segments except intracardiac lead, which could not be evaluated, as there were no true negative intracardiac lead controls because none of the control patients underwent myocardial uptake suppression protocol.

Spread of the infection from a contaminated generator pocket through the subcutaneous lead into the endovascular space has been hypothesized to be the main pathogenic mechanism in CIED infections.4 This mechanism may explain 83% (34/41) of our SI cases. In addition, in our data, the [18F]FDG-PET/CT CIED pocket was the most frequent area of positive uptake, followed by subcutaneous lead. Nonetheless, Rizwan et al.10 suggests that CIED lead infection can also originate from a distant source, possibly explaining the remaining 7 cases (17%) with SI but without LI.

Compared with previous studies, our work shows equivalent sensitivity and specificity values with a larger sample of patients. In our cohort, the ROC curve for pocket SUVmax had a cutoff point of 2.4 with sensitivity of 79.6% and specificity of 92.6% (figure 2A). Other studies have reported similar results for diagnostic yield in pocket CIED infections.12–15 In contrast, Mahmood et al.7 showed higher sensitivity and specificity values for SI, probably due to a meta-analysis based on several heterogeneous studies with a small number of patients, divergent designs, and the inclusion of other prosthetic infections.

Eight out of 47 cases with LI showed normal [18F]FDG-PET/CT results considered as false negatives. In all but 1 false negative result, the patients had undergone antibiotic therapy for more than 20 days before [18F]FDG-PET/CT acquisition. Several studies have suggested that antibiotic therapy for more than 7 days before [18F]FDG-PET/CT acquisition can reduce its diagnostic performance.11,12,16 However, no significant differences were found in our cohort in the period between antibiotic initiation and [18F]FDG-PET/CT performance (a median of 13 days for false negatives and 5 days for true positives, P=.19). Nonetheless, significance could be masked by the small number of cases. The absence of false positive results in our cohort can be partially explained by the longer period between device implantation and [18F]FDG-PET/CT acquisition in controls, which was a median of 6.1 [0.05-24.31] years. In the study by Jeronimo et al.,12 the median time between device implantation and [18F]FDG-PET/CT was 2.3 [0.6-6.4] years. That study, as well other published works,14,15 state that false positive results are caused by postoperative inflammatory activity.

Although TEE plays an essential role in the diagnosis of lead infection, it may be hard to differentiate vegetations from lead strands or small adhered thrombi.16 It is commonly accepted that TEE is initially performed in patients with suspected SI, whereas [18]FDG-PET/TC should be the primary technique to confirm LI due to the lower sensitivity of [18F]FDG-PET/CT for endovascular and intracardiac lead infections. Concordantly, in our cohort, TEE showed higher accuracy in diagnosing intracardiac lead infections. However, it is worth noting that the performance of [18F]FDG-PET/CT was better in subcutaneous and endovascular lead infections in SI cases with bacteremia. A negative TEE result does not rule out SI 12 and, considering that Pizzi et al. demonstrated an increased sensitivity of [18F]FDG-PET/CT in combination with TEE,17 our results suggest that [18F]FDG-PET/CT may not be the only the test of choice to confirm an active local infection15 but may also be complementary to TEE in SI cases. Our data showed that [18F]FDG-PET/CT used in combination with TEE significantly increased the rate of definite diagnosis of infection from 30.4% to 56.1% (P=.04) due to the detection of endovascular lead [18]FDG uptake. Furthermore, [18F]FDG-PET/CT has the additional value of being able to detect septic embolisms,14,18–20 as occurred in 2 of our SI cases. This datum seems to be consistent with that published by Rodríguez-Alfonso et al.,21 who showed that [18F]FDG-PET/CT correctly reclassified 57% of patients with initial suspicion of generator pocket infection by detecting lead infection with high diagnostic performance, especially in patients with initial suspicion of LI.

Some authors suggest that an increase in the metabolic rate of the spleen and BM could be used as an indirect sign of infection.4 Our study could not corroborate this hypothesis, as SUVmean spleen and SUVmean BM were similar in cases and controls and between LI and SI. However, most control cases were patients with cancer, in whom spleen and/or BM uptake could have been increased due to their neoplastic disease, chemotherapy, or other hematological alterations. Nonetheless, we found significative differences in spleen and BM metabolism between those patients with SI and confirmed bacteremia compared with LI cases. These results may be explained by the expected hyperactivation of the phagocytic mononuclear system in cases of bacteremia, which could be helpful in distinguishing bacteremic lead infections from isolated LI.

Complete device removal in CIED-IE is mandatory to cure infection4,22; however, in the last few decades a higher number of patients cannot undergo complete CIED extraction surgery,5 even if indicated, due to the growth in comorbidities, older age, and more complex infections. In these cases, CAS has been proposed as a helpful strategy. In our cohort, patients with incomplete device removal received undefined CAS, usually lifelong, which represented a heavy burden for patients and led to adverse effects, multidrug-resistant infections, and a high cost for the health system. To date there is no tool to guide clinicians on when to stop CAS. We studied 6 cases in which [18F]FDG-PET/CT, in combination with the clinical course and laboratory and microbiological findings usefully guided physicians in discontinuing CAS in the absence of relapse for more than 2 years of follow-up. Despite the limited number of cases in our cohort, this study supports the idea that further prospective studies could validate [18F]FDG-PET/CT as a reliable tool for discontinuing CAS safely during the follow-up of cases with incomplete device removal.23,24

The diagnostic performance of [18F]FDG-PET/CT is high in local CIED infections but lower in endovascular and intracardiac lead infections. However, [18F]FDG-PET/CT is the only available technique for assessing subcutaneous and endovascular lead infection and may be complementary to TEE in cases of bacteremia, increasing the definite diagnosis of lead infections. Moreover, spleen and BM metabolism may help to distinguish between bacteremic lead infections and isolated LI. Although further prospective studies are needed, follow-up [18F]FDG-PET/CT could potentially play a role in the management of CAS therapy when complete device removal is unachievable.

J. M. Miró has an Instituto de Investigaciones Biomédicas August Pi i Sunyer (IDIBAPS) personal intensification research grant from 2017 to 2023. MHM held a Rio Hortega Research Grant (CM17/00062) from 2018 to 2020. All other authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

All the authors contributed to the conception and design, data acquisition, drafting of the article, critical revision, and final approval of the manuscript. The data underlying this article will be shared on reasonable request to the corresponding author (data available on request). M. Hernández-Meneses and A. Perissinotti contributed equally as first authors. The members of Hospital Clinic of Barcelona Infective endocarditis Team investigators are listed in the supplementary data.

None of the authors have any association that might pose a conflict of interest in this work. J.M. Miró, as the corresponding author, declares having no conflicts of interest. As a corresponding alternative author, D. Fuster claims to have no conflicts of interest.