Author + information
- Published online December 3, 2018.
- Cristina Sánchez-Enrique, MD,
- Carmen Olmos, MD, PhD∗ (, )
- Ana Jiménez-Ballvé, MD, PhD,
- Cristina Fernández-Pérez, MD, PhD,
- Carlos Ferrera, MD, PhD,
- María Jesús Pérez-Castejón, MD, PhD,
- Aida Ortega Candil, MD, PhD,
- Roberto Delgado-Bolton, MD, PhD,
- Manuel Carnero, MD, PhD,
- Luis Maroto, MD, PhD,
- José Luis Carreras, MD, PhD and
- Isidre Vilacosta, MD, PhD
- ↵∗Instituto Cardiovascular, Hospital Clínico San Carlos, Care of Profesor Martín Lagos s/n, Madrid 28040, Spain
Positron emission tomography/computed tomography (PET/CT) is helpful in prosthetic valve (PV) infective endocarditis (IE), but its usefulness in other settings is less established (1). We aim to evaluate the yield of PET/CT in patients with suspicion of IE in 3 scenarios: PV, native valves (NV), and cardiac implantable electronic devices (CIED).
Of 120 patients with suspicion of IE prospectively recruited, 95 with 139 potential cardiac foci of infection (PFI) (77 PV, 34 NV, 28 CIED) were systematically studied with transesophageal echocardiography (TEE), blood cultures, and PET/CT. In 25 patients, PET/CT was not performed due to hemodynamic instability or need for emergent surgery.
PET/CT analysis was made based on visual interpretation, considering abnormal (positive) the presence of focal or heterogeneous hypermetabolic intensities that persisted in the noncorrected images. Extracardiac uptakes indicating embolisms were considered a minor criterion for IE.
Modified Duke criteria (DC) were applied to each PFI, that is, predisposing heart conditions. Histology in patients who underwent surgery or necropsy was considered the gold standard of IE. When anatomy was not possible, gold standard was established by the “Endocarditis Team” at 6-month follow-up, taking into account blood cultures, TEE, PET/CT, and other potential sources of infection.
The diagnostic performance of DC and PET/CT for each PFI was compared with the gold standard. The performance of DC when adding PET/CT findings (major and minor criteria), according to the last IE guidelines (1) was also evaluated. Sensitivity (Se), specificity (Sp), and predictive values were calculated using the final diagnosis of confirmed or rejected IE.
PET/CT was positive in 73 PFI. Agreement on PET/CT interpretation between 2 independent readers was high (kappa: 0.7).
Forty-one patients with 57 PFI underwent surgery. Twelve patients died, and necropsy was performed in 3 with 5 PFI. Accordingly, 62 of 139 PFI were anatomically studied (44.6%), and IE was documented in 45. Other 25 foci were considered infected by consensus. Thus, active IE was established in 70 foci.
Se and Sp of PET/CT were, respectively, 97.5% and 56.8% in PV, 47.6% and 84.6% in NV, and 55.6% and 94.7% in CIED. Echocardiography and PET/CT results were concordant in 62.5% of foci; among cases with disagreement, PET/CT established the diagnosis of infection in 7 and ruled it out in 15.
In suspected PV and CIED IE, the addition of PET/CT results to DC increased Se, from 87.5% to 97.5% in PV, and from 55.6% to 77.8% in CIED. Sp was reduced from 91.9% to 67.6% in PV and was not modified in CIED (84.2%). In suspected NV IE, the addition of PET/CT to DC did modify neither Se (61.9%) nor Sp (92.3%).
The best PET/CT negative predictive value (95.4%) was obtained in PV cases, in which IE was finally ruled out in 21 of 22 cases with negative studies (Figure 1). However, 16 patients with PV in whom IE was ruled out had a positive PET/CT.
In CIED, PET/CT was particularly accurate in possible IE, as PET/CT results coincided with the final diagnostic consensus in 8 of 11 cases.
In NV, the Se of TEE was much greater than that of PET/CT (95% vs. 47.6%). However, the specificity of PET/CT was satisfactory (84.6%). Thus, PET/CT may still be useful in patients with suspicion of NV IE when echocardiography is doubtful. Interestingly, in patients in whom PET/CT was positive, the bulk of infection was greater.
Compared with previous studies (2), in the current work, DC had better diagnostic accuracy. This might be related to the gold standard used, as well as a better performance of TEE in our study.
False positives in PV may result from chronic inflammation, use of surgical adhesive (3), and increased 18F fluorodeoxyglucose uptake of the normal myocardium.
This study has limitations. A selection bias may be present, as we mostly had anatomic information from patients considered positive and sent to surgery. PET/CT was not performed in patients who were hemodynamically unstable. The sample size for NV and CIED is rather limited to draw definitive conclusions. Lastly, the Endocarditis Team was not blind to PET/CT results.
In conclusion, PET/CT is especially useful in patients with suspicion of IE who do not meet definite DC. Our results support the use of this technique in PV and CIED, but it is less helpful in NV. Sp of PET/CT in clinical practice may be lower than has been published previously. A multicenter study including a larger number of patients with NV and CIED is needed.
Please note: The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- 2018 American College of Cardiology Foundation
- Pizzi M.,
- Roque A.,
- Fernández-Hidalgo N.,
- et al.
- Mathieu C.,
- Mikaïl N.,
- Benali K.,
- et al.