A retrospective analysis was conducted of patients evaluated by PET/CT for suspected CDRI from August 2011 to January 2018. Diagnosis had been previously established according to local and systemic signs or symptoms of infection, microbiological findings (blood cultures or initial generator pocket exudate culture), and echocardiographic (transthoracic and transesophageal) findings.

Patients were classified by applying the diagnostic categories proposed by Sandoe et al.12 (table 1) at 3 time points: according to conventional diagnostic tests (pre-PET diagnosis), after including the results of PET/CT (post-PET diagnosis), and according to the reference standard (definitive diagnosis). We added 2 more categories, fever of unknown origin (FUO) and bacteremia without microbiological criteria of endocarditis according to modified Duke criteria (non-Duke bacteremia), because these categories constituted a reason for PET/CT for suspected CDRI in our setting. For post-PET diagnostic classification, in the event of a mismatch between echocardiography and PET/CT, the results of the latter technique were accepted. The consensus decision of the multidisciplinary endocarditis team was considered as the reference standard after a review of all the results, including those of the exudate culture and the follow-up data obtained at consultations 10 to 14 days after discharge or at the end of antibiotic therapy, with new standard blood cultures and surveillance of signs of infection. In addition, the team also included information obtained from a review of medical records over the first year in search of consultations or readmissions due to suspected infection.

The microbiological samples were incubated for 14 days and the extracted CIED components underwent sonication. The study was approved by the Research Ethics Committee.

The PET/CT acquisition protocols were followed in accordance with the guideline recommendations.9–11 We applied a protocol for the suppression of the myocardial physiological uptake of 18F-FDG, which consisted of prolonged fasting (16hours) and, since 2016, an intravenous bolus (i.v.) of sodium heparin (50 IU/kg) 15minutes before injection of the radiotracer. After confirming that blood glucose levels were less than or equal to 200mg/dL, we administered a standard dose of 18F-FDG 370 MBq i.v. After 60minutes, images were acquired from the cranial vertex to the upper third of the thighs using a Biograph 6 PET/CT device (Siemens, Germany).

The images underwent qualitative analysis using Leonardo workstations with Syngo 2002A_R1.0 imaging software and Syngo.via (Siemans, Germany). Focal or heterogeneous uptake superior to the vascular pool or healthy tissue adjacent to the generator pocket (local infection), leads, or prosthetic valve (IE) were considered suggestive of infection. We also evaluated findings compatible with septic embolism, infectious/inflammatory disease unrelated to CIEDs, or tumora disease.

For diagnostic performance analysis, we used reference standard contingency tables to calculate the sensitivity, specificity, positive and negative predictive values, and positive and negative odds ratios of echocardiography and PET/CT.

To be able to construct the contingency tables, the definitive diagnosis categories agreed upon as the reference standard were as follows:

• •

  No IE: superficial inflammation, uncomplicated generator pocket infection, FUO, and non-Duke bacteremia.

• •

  IE: complicated generator pocket infection, possible or definitive lead infection, and definitive CIED-IE/prosthetic valve infection.

The following groups were established for the imaging tests:

• •

  IE: mobile echogenic mass (echocardiography) or pathological uptake (PET/CT) in leads or endocardial surface.

• •

  No IE: due to absence of vegetation (echocardiography) or lack of uptake or uptake limited to the generator pocket or unrelated to CIED (PET/CT).

Uncertain results on imaging tests were considered positive, favoring S, due to the life-threatening nature of the process under study. A second analysis was conducted in which uncertain results were considered negative, while taking into account that this approach would be less acceptable in clinical practice.

PET/CT performance was calculated based on the assessment of isolated generator pocket infection or deep generator pocket infection, and the following groups were established:

• •

  Generator pocket infection: uncomplicated and complicated generator pocket infection.

• •

  Nongenerator pocket infection: superficial inflammation, possible or definitive lead infection, definitive CIED-IE/prosthetic valve infection, FUO, and non-Duke bacteremia.

The weighted kappa index was used to calculate the agreement between pre-PET diagnosis and definitive diagnosis, and between post-PET diagnosis and definitive diagnosis in relation to the distribution of patients in the diagnostic categories. To this end, the categories of possible or definitive lead infection and definitive CIED-IE/prosthetic valve infection were grouped as “CIED-IE/prosthetic valve infection” and the categories of FUO and non-Duke bacteremia were grouped as “no CDRI”. Applying this grouping, the percentage of correct diagnostic change within each pre-PET diagnostic category was analyzed after including the PET/CT results to identify a clinical subgroup deriving the greatest benefit from this imaging modality.

We used the SPSS (Version 22.0) statistical software package and calculators to evaluate diagnostic tests conducted by the Ramón y Cajal Hospital Research Unit (Madrid, Spain).13

We included 44 patients. Table 2 shows their main characteristics. Most CIEDs (79%) had at least 2 leads. All generators were implanted in the left anterior thoracic wall. The initial culture of the pocket region was taken if there was lesion exudate. In this case, the swab sample was taken directly from the wound.

Table 3 shows the tests that were performed and patient classification according to the results. Table 4 shows details of blood cultures according to the definitive diagnosis.

In 25 patients (57%), IE was established according to the reference standard, which was higher among patients referred with local infection (64% vs 53%).

Table 2 and table 3 show the results of CIED extraction and culture in 19 patients (43%). CIED extraction was more common among patients with a definitive diagnosis of IE than among those with a definitive diagnosis of uncomplicated generator pocket infection or no CDRI (61.5% vs 10.5%; P = .01).

Treatment was conservative in 25 patients (57%): of these, 7 had a definitive diagnosis of IE and high surgical risk and were treated with antibiotic therapy i.v. for 4 to 6 weeks; 4 were treated with chronic oral antibiotic therapy and antibiotic therapy i.v. for 6 weeks; and 3 had prosthetic valve IE unrelated to the CIED. During follow-up, infection was controlled in 24 patients: however, 1 patient died at 8 months due to sepsis of intestinal origin. The eighth patient with a definitive diagnosis of complicated infection was initially treated as having uncomplicated infection (oral antibiotic therapy for 4 weeks). Unfavorable clinical course led to CIED extraction some months later.

The 5 patients with a definitive diagnosis of generator pocket infection were treated with oral antibiotic therapy for 4 weeks, without relapse of infection during follow-up. CDRI was ruled out in 14 patients (32%) and a specific alternative diagnosis was reached in 9 of them (table 3).

Table 3 shows the findings of echocardiography and PET/CT. All patients underwent at least 1 echocardiography (34 transesophageal, 34 transthoracic, and 25 both). In case of disagreement, the result of the transesophageal approach was accepted. Echocardiographic examination was considered positive in 9 patients (20%), uncertain in 6 (14%), and negative in the other patients.

The results of PET/CT and echocardiography were concordant in 19 patients (43%): 13 patients without findings and 6 patients with findings of IE. Table 5 shows the diagnostic performance of PET/CT and echocardiography. The sensitivity of PET/CT was significantly higher than that of echocardiography, and showed pathological uptake in 15 of the 16 false negatives (FN) on echocardiography. Uptake was located in the external course of the leads in 14 patients and in the aortic valve in 1 patient.

The specificity of PET/CT was higher than that of echocardiography without reaching statistical significance. None of the 7 patients with false positive (FP) findings on echocardiography showed pathological uptake. The results of PET/CT are supported by the finding that the disease was resolved after conservative treatment not directed at IE in 5 patients and the absence of infection during and after CIED extraction in the other 2 patients.

PET/CT showed 1 FP result for IE consisting of pathological uptake in a proximal lead of a bicameral CIED that had been replaced 35 days earlier. Oral antibiotic therapy for 1 month was administered for local infection and negative echocardiogram. After treatment, PET/CT showed persistent uptake, although of decreased intensity. However, there were no new infectious episodes during the year of follow-up, and the definitive diagnosis was uncomplicated CIED infection.

Four FNs on PET/CT were detected, 3 in patients with systemic infection, positive findings on echocardiography, and positive blood cultures. Because of the risk of CIED extraction, they were treated with antibiotic therapy i.v. for 6 weeks. The team established a definitive diagnosis of suspected lead infection in 2 of these patients. Due to severe heart failure, the third patient received a transplant after initial antibiotic therapy; pretransplant echocardiography showed an image compatible with treated vegetation. The fourth patient showed local infection and negative findings on echocardiography; no blood cultures were taken. Superficial infection was suspected and oral antibiotic therapy was initiated. An unfavorable response led to CIED extraction; Staphylococcus epidermidis infection was confirmed by generator and lead culture. Myocardial suppression was optimal in 3 patients and suboptimal in 1. Antibiotic therapy time prior to PET/CT was 0, 1, 4, and 8 days, respectively.

Table 6 shows the diagnostic performance of echocardiography and PET/CT according to clinical presentation. The sensitivity of PET/CT was particularly high in patients with local symptoms. In patients with systemic symptoms, a correlation was found between pathological uptake in the deep components of the CIED or in the endocardial surface and the definitive diagnosis of IE.

Table 3 and table 5 show the performance of PET/CT in generator pocket infection. The only FP was associated with a patient with systemic infection undergoing cardiac transplant due to hypertrophic cardiomyopathy with incomplete CIED removal 3 months earlier. Because of intermittent bacteremia due to Pseudomonas aeruginosa, PET/CT was performed showing uptake in the lead retained in the vena cava and aortic root. P. aeruginosa infection was confirmed by aortic root and lead culture following aortic root replacement and lead extraction. PET/CT also showed mild uptake in the generator pocket site. Infection was finally attributed to postoperative inflammation in the absence of other clinical or microbiological data confirming local infection.

Two of the 3 patients with FN on PET/CT had local infection. Treatment was conservative with an unfavorable clinical course. The CIEDs were extracted and culture performed, which confirmed uncomplicated and complicated CDRI, respectively. The third patient with FN on PET/CT had local infection and heart failure. After 2 weeks of oral antibiotic therapy with local improvement, the definitive diagnosis was assumed to be uncomplicated generator pocket infection. None of the patients had been started on antibiotic therapy before the PET/CT study.

Table 7 shows the distribution of patients according to the 3 diagnostic steps. Concordance between the post-PET diagnosis and the definitive diagnosis was excellent (κ=0.81) and concordance between the pre-PET diagnosis and definitive diagnosis was low (κ=0.36)

Using the model of Graziosi et al.,14figure 1 shows individual changes between the initial and final diagnostic classification according to the results of PET/CT and the reference standard. The most promising PET/CT results were obtained in patients who had been initially referred for uncomplicated generator pocket infection, because 8 of 14 patients (57%) were correctly reclassified by the detection of pathological uptake in leads, indicating complicated infection. In 7 of 21 patients (33%) with suspected deep infection, the absence of pathological uptake contributed to its correct exclusion. Among the 9 patients referred for FUO or non-Duke bacteremia, PET/CT detected deep infection in 2 patients and increased the level of certainty in ruling out CDRI and provided alternative diagnoses in 4 patients.

Figure 1.

Individual changes between the initial and final diagnostic classification according to the results of PET/CT. CIED-IE: infective endocarditis associated with an intracardiac device; IE: infective endocarditis; FN: false negative; FP: false positive; PET/CT, positron emission tomography-computed tomography.

(0.35MB).

In 6 patients (14%), PET/CT showed uptake in the deep components of the CIED as well as pulmonary uptake indicative of embolic origin. In 2 patients (4%) with lead infection and bacteremia, the technique detected concomitant osteoarticular infection (figure 2). In a previously described patient, it detected an unsuspected aortic root infection.

Figure 2.

Pacemaker patient, 84 years, last manipulation 3 years earlier. PET/CT was performed for fever and methicillin-sensitive Staphylococcus aureus bacteremia. A: volumetric imaging. B: lead uptake in upper cava according to PET/CT fusion imaging (arrow). C: an unsuspected septic focus on the right hip with periarticular fluid collections (asterisk). Considering the baseline situation, conservative treatment was administered consisting of cefazolin i.v. and oral rifampicin for 6 weeks. Blood cultures were negative on day 5. This treatment was followed by oral cephalosporin indefinitely without new episodes of infectious exacerbation during the subsequent year. PET/CT, positron emission tomography-computed tomography.

(0.19MB).

In 4 of the 14 patients (29%) in whom the definitive diagnosis ruled out CDRI, PET/CT facilitated an alternative diagnosis by the detection of interstitial pneumonitis, active pneumonia, papillary thyroid carcinoma, and pericarditis (figure 3).

Figure 3.

Patient, 76 years, with implantable automatic defibrillator, last manipulation 3 years earlier. PET/CT performed for general malaise, fever of unknown origin, and negative blood cultures showing pericardial pathological uptake. A: PET transaxial image. B: PET/CT transaxial image (asterisk). C and D: absence of lead uptake (coronal images). A diagnosis of pericarditis was made. Colchicine treatment was initiated, with clinical remission. PET/CT, positron emission tomography-computed tomography.

(0.19MB).

In 4 patients (9%) intestinal uptake was detected and confirmed as low-grade tubular adenoma in 2 patients and high-grade neoplasia in 1 patient. Complementary studies could not be performed in the fourth patient due to early death.

This study has several limitations. First, its main limitation is its retrospective design, which makes it difficult to identify and control confounding factors, thus limiting the extrapolation of the results. Second, selection bias may be present, because patients with clear surgical indications or in a severe or unstable clinical situation may not have been referred to PET/CT. Such bias would most likely affect the results of echocardiography in our series, which therefore would not reflect actual echocardiographic performance in patients with CIEDs. In our study, a broad spectrum of patients with systemic infection were referred to PET/CT and therefore IE was one of the more likely diagnostic outcomes. In contrast, patients with clear generator pocket infection are more likely to have infection of the deep components of the CIED. This would explain why, in our series, the percentage of IE was higher in patients referred for local symptoms. Finally, PET/CT images were analyzed when the results of the other tests were already known. Likewise, the endocarditis team established the definitive diagnosis after knowing the result of the PET/CT, which may have influenced their decisions. In both cases there may have been diagnostic bias.