Infectious endocarditis: Improving Diagnostic Performance in Native and Prosthetic Valve Infection with 18F-FDG PET/CT

The diagnosis of infective endocarditis (IE) remains a clinical challenge. Diagnostic accuracy of the modied Duke criteria is suboptimal for native valve endocarditis (NVE) and even worse in the presence of prosthetic material-related infection (PVE). We aim to evaluate the diagnostic performance of 18F-FDG PET in patients with suspected IE referred to perform PET/CT. Methods: Consecutive patients with suspected IE, referred to perform PET/CT between May 2016 and June 2019 were included. Diagnostic performance of modied Duke criteria (mDC) and PET/ CT for IE for NVE and PVE was determined. Results: In total, 82 patients (mean age of 61 ± 19 years, 62% of male gender) were enrolled. There were 67 18F-FDG PET/CT concordant results with nal diagnosis, corresponding to a 96% of agreement, k=0.91(p=0.04). A SUVmax cutoff value of ≥ 3.1 identied positive cases with 88.9% sensitivity and 70.0% specicity. In patients with NVE, the addition of PET/CT to the mDC resulted in a reduction of the number of possible IE cases (from 58% to 4.3%). In patients with PVE/intracardiac devices, PET/CT allowed reclassication of 67.4% of possible cases to 4.2%. An alternative diagnosis was provided in 55.6% of the negative IE cases. Conclusions: 18F-FDG PET/CT proved to be a useful diagnostic tool in patients with both suspected NVE and PVE with good sensitivity and specicity, resulting in a signicant decrease of the number of possible endocarditis. Furthermore, it allowed the identication of the cause of clinical scenario in more than 50% of patients in which the diagnosis was excluded.


Introduction
The diagnosis of infective endocarditis (IE) is a clinical challenge, due to a myriad of nonspeci c clinical presentations, involvement of different microorganisms and patient characteristics. Despite ongoing advances in both diagnostic and therapeutic procedures, IE remains associated with a poor prognosis, with an in-hospital mortality of 14-22% and 1-year mortality of 40%(1) (2). Modi ed Duke criteria, the cornerstone of its diagnosis, is largely based in the presence of positive blood cultures and evidence of endocardial involvement on echocardiography. However, both transthoracic (TT) and transoesophageal echocardiography (TEE) miss IE sequelae in 30% of patients, especially, when intracardiac prosthetic material is implied (3). Therefore, the sensitivity and speci city of the modi ed Duke criteria is, approximately, 80% for native valve endocarditis (NVE) and lower values were found for prosthetic (PVE) or implantable cardiac electronic device (ICEDs), which leads to an erroneous diagnose in several patients (1).
18F-FDG PET/CT, combining the high sensitivity of 18F-FDG PET to detect in ammation with the high spatial resolution of cardiac CT to de ne structural damage, is evolving as an important additional method in di cult-to-diagnose cases and is able to provide functional data on the extent of IE. The added diagnostic value of 18F-FDG PET/CT has been demonstrated in patients suspected of PVE or infections related to de brillators, pacemakers, or ventricular assist devices (4). For extracardiac infection, 18F-FDG PET/CT detects a signi cant number of clinically relevant foci in patients with known or suspected NVE or PVE (1)(5). In the most recent European Society of Cardiology (ESC) guidelines, 18F-FDG PET/CT has been included in the diagnostic algorithm PVE and also in the detection of embolic events for both NVE and PVE (6)(7).
We aimed to evaluate the diagnostic performance of 18F-FDG PET in a cohort of patients with suspected IE referred to PET/CT in a tertiary care hospital.

Study population
A retrospective study was performed at a tertiary center with 18F-FDG PET/CT and included all referred patients for this exam for suspected IE between May 2016 and June 2019. The choice to perform 18F-FDG PET/CT and the IE suspicion were based on the attending endocarditis team and did not follow a standardized protocol.
Clinical, laboratory and microbiological data Baseline demographic characteristics of patients, including all relevant clinical data, were collected from hospital records at hospital admission. Laboratory data included basic renal and liver function test as well as hemogram. For microbiological data, all blood cultures with respective antibiotical sensitivity testing and, when applicable, serology, were recorded.
Other imaging tests data Transthoracic echocardiography was performed in all patients and transesophageal echocardiography performed in 86.3%, and recorded images and reports were analyzed. Data was extracted regarding assessments of general cardiac parameters, including cavity dimensions and ventricular function, and speci c endocarditis suggestive ndings such as valve function, vegetations (shape, size, and number), and periannular complications (abscess, pseudoaneurysm, perivalvular leak).

PET/CT
Whole-body scans and dedicated cardiac bed images were obtained in a hybrid PET/CT scanner (Discovery ST, General ElectricsÒ). A [18F] FDG myocardial uptake suppression protocol, that included a 48h high-fat and low-carbohydrate diet and a fasting period of at least 12 hours, was used to optimize image quality and diagnostic accuracy. To further improve image quality 50 IU/kg of unfractionated heparin was administered, 15 minutes prior to radiotracer injection, in a few patients. Blood glucose levels were required to be less than 180 mg/dL during a period of approximately 60 min before the administration of the 18F-FDG.
Attenuation corrected images were used to locate and examine the hypermetabolic foci. However, nonattenuation corrected images were also reviewed, particularly to assess hypermetabolic foci near metallic structures, mainly valvular prosthesis, medical devices and electrodes. Images were reviewed by both a nuclear medicine physician and a cardiologist with several years of experience in cardiac [18F] FDG PET/CT. Uptake intensity was measured by means of maximum Standardized Uptake Values (SUVmax) in patients with and without ndings suggestive of infection. Uptake patterns were also noted. Cardiac and extra-cardiac ndings were reported. Extra-cardiac ndings included both those attributable to IE, such as embolic events, but also indicative of other pathologies, such as neoplasic lesions.

Final diagnosis
The nal diagnosis of IE (gold standard) was established by consulting the nal diagnosis attributed to the patient by the Endocarditis team at the time of hospital discharge or death, after being possession of clinical, microbiological, and imaging information as well as clinical response to treatment.

Statistical analysis
Continuous data are presented as the mean +/-SD. Categorical data are expressed as percentages.
Sensitivity, speci city, positive and negative predictive values of DC at hospital admission, TOE echo and PET/ CT, in evaluation of NVE and PVE were estimated. Receiver-operating characteristic (ROC) curves were built to determine the cutoff values with the best sensitivity-speci city combination to detect infection considering a binary outcome (1=positive 0=negative). Continuous data was tested for normality with the Shapiro-Wilks test and the Student's t or the Mann-Whitney test was applied to compare them as appropriate. Fisher exact test was used for nominal variables to assess differences between groups. In these analyses we used the nal diagnosis of con rmed and the rejected NVE and PVE as the gold standard outcome. P values of less than 0.05 were considered statistically signi cant.

Results:
Clinical and microbiological data and echocardiography ndings In total, 82 patients (mean age of 61 +/-19 years, 62% of male gender) were enrolled from May 2016 to June 2019. Baseline characteristics were summarized in Table 1.
Half of included patients had an implanted prosthetic valve (n=35), the majority of which biological (n=19) and in aortic position (n=27). Fever was present in most patients (n=64) ( Table 2), followed by heart failure symptoms and vascular occurrences. One patient presented with pacemaker pocket signs of infection. Staphylococcus aureus was identi ed in 19 cases and more than half of these were methicillin resistant (52.9%). Presence of vegetations was the most common echocardiographic nding (n=31), followed by moderate to severe regurgitation (n=14), presence of abscess, pseudoaneurysm (n=9) or stulae (n=5).
PET/CT ndings Seventy patients (90.9%) were under antibiotic therapy before PET/CT with a mean interval between treatment initiation and PET/ CT of 20±14 days. PET/CT was positive for IE in 24 cases. The receiveroperating characteristic curve yielded an area under the curve of 0.91 (95% con dence interval, 0.88-0.96) for SUVmax. A SUVmax cutoff value of ≥3.1 identi ed positive cases with 88.9%% sensitivity and 70.0% speci city (Figure 1).

PET/ CT performance in IE of NVE and PVE diagnosis
There were 6718F-FDG PET/CT concordant results with the nal diagnosis of the endocarditis team, corresponding to a 96% of agreement, k = 0.91(p=0.04).
PET/CT was positive in 1 patient without a nal diagnosis of IE (false positive, 2.2%). Conversely, PET/CT was negative in 4 patients with a diagnosis IE (false negative, 5.7%).
The sensitivity, speci city, positive, negative predictive values and area under the curve (95% con dence intervals) of DC at admission, echocardiogram and PET/CT in the diagnosis of IE are shown in Table  3.The addition of PET/CT to the modi ed DC resulted in a substantial increase in diagnostic speci city (from 33.3% to 97.8%) with only a mild reduction in sensitivity (from 88.0% to 84.0%). The percentage of IE cases classi ed as "possible" was reduced from 58% to 4.3%.
Concerning the subgroup of patients with intracardiac devices, PET/CT showed similar sensitivity to DC at admission but superior to echocardiography (89.5% vs 68.4%). PET/CT was also more speci c than echocardiogram and DC at admission for diagnosis of device IE (90.0% vs 60.0% vs 51.0%). In this subgroup, PET/CT allowed reclassi cation, reducing from 67.4% of possible cases to 4.

Final Diagnosis and Patient Management
The patients were followed during a mean of 14 +/-10months. According to diagnosis gold standard de ned for this study, IE was established in 22 patients and was rejected in 45 patients; 3 patients were classi ed as possible IE. Those patients with de nitive diagnosis of IE, 11 were submitted to surgery or percutaneous extraction of leads or devices. The remaining patients were medically managed. The allcause mortality was 17.1% (12 patients).

Discussion:
Our study nds that the use of PET/CT adds further diagnostic information to classical DC in patients with suspected IE, particularly those with prosthetic valves and intracardiac devices. This may have a signi cant effect on the choice of an appropriate strategy and therefore in uence IE-related morbidity and mortality.
Infective endocarditis is more common now than in the past, with its incidence increasing from 9.3 per 100 000 population in 1998 to 15 per 100 000 in 2011 (8). Partially this is due to health care-associated disease (9). In a large multicenter, multinational study, health care-associated infective endocarditis accounted for 34% of cases (10). Hemodialysis, non-hemodialysis intravascular catheters, and invasive procedures are often associated with the infection (11). Furthermore, the proportion of cases related with prosthetic valves and implantable cardiac devices is increasing (8). Despite major advances in diagnostic and therapeutic procedures, the prognosis is poor with a 1-year mortality approaching 30% and high complication rates at long term(1)(12).
Patients with intracardiac devices or prosthetic valves are a special IE population with a clear predisposing factor, different epidemiological pro le, and higher mortality. The traditional modi ed Duke criteria are di cult to use in these patients due to the challenging interpretation of lesions on echocardiography and several cases of suspected IE are left without a conclusive diagnosis(2). The scenario is even more problematic in case of suspected intracardiac devices/leads infection, in which patients frequently present with nonspeci c clinical manifestations and both TT and TOE were limited in evaluation of cardiac right chambers. Consequently, these patients may be recurrently hospitalized for an in ammatory/infectious disorder of unknown origin despite detailed investigations. In the absence of proven lead infection, further clinical management including the removal of the system is postponed due to the known morbidity and mortality risks associated with the extraction procedure (0.5-2.0 %) (13).
The 2015 ESC Guidelines on Endocarditis recommend using additional imaging modalities when echocardiography and blood cultures are inconclusive (i.e. result in a 'possible' diagnosis of endocarditis, or a 'rejected' diagnosis with persisting high suspicion). Three techniques may be employed: CTA to depict perivalvular complications, cerebral magnetic resonance imaging (MRI), and/or whole-body SPECT/CT or PET/ CT to depict embolic events and evaluate abnormal metabolic activity around the site of prosthetic valve implantation (14).
In a cohort of patients with suspected IE referred to PET/CT in a tertiary care hospital the its diagnostic performance was evaluated. Comparing with modi ed DC the use of PET increased signi catively the speci city for the diagnosis of IE from 33 to 98% in NVE and from 15.8 to 94.5% in PVE with similar sensitivity.
In a metanalysis, Maryam Mahmood et al including 13 studies involving 573 patients showed, a pooled sensitivity of PET/CT for diagnosis of IE of 76.8% and the pooled speci city was 77.9%. Diagnostic accuracy was improved for PVE with sensitivity of 80.5% and speci city of 73.1% (15). However, some of these studies included initially rejected patients by modi ed Duke criteria, probably making less costeffective the indication of 18F-FDG PET/CT in suspected IE (5). Our study did not include initially rejected IE patients by modi ed Duke criteria showing a high speci city in all the indications in which 18F-FDG PET/CT was performed. Furthermore, the results of this study indicate that substantial bene ts can be obtained by including PET/CT in the diagnostic workup of patients with both native and with prosthetic valves/ intracardiac devices suspected IE.
No distinction between biological and mechanical prosthetic valves was made in our study. Roque et al., in a study evaluating metabolic patterns of captation after surgery, no differences between biological or mechanical prothesis were found (16). Evidence of its use in patients suspected of transcatheter-replaced aortic valves (TAVR) endocarditis is still limited to case reports, but may soon become relevant as the incidence of TAVR endocarditis has been increasing over the past decade and the mortality is high. In our cohort, 2 patients had TAVR endocarditis suspicion and 1 patient had a previous "mitraclip" implantation. In all cases, PET-FDG excluded the diagnosis.
The 4 false-negative cases, in our study, could be due previous antibiotherapy before PET/CT; and presence of small lesions below the metabolic/spatial resolution of PET/CT. In fact, both false negative and false positive results have been reported in previous studies. False negative results might be due to prior administration of antimicrobial therapy, small size of vegetations, and elevated blood glucose concentration. False positive results might be a result of recent cardiac procedures and inadequate patient preparation (17).
There are other valid reasons to consider both PET/CT even when the Modi ed Duke criteria have already been met, which include: evaluation of involvement of other valves or cardiac implanted electronic devices, identi cation of port of entry, identi cation of other foci of infection or other cause for clinical picture, all of which may guide treatment strategies (1).The presence of septic emboli is crucial for the correct management of patients with IE. Failure to identify metastatic infection complications may lead to early interruption of therapy, thus potentially triggering relapse and an unfavorable outcome. (18) Infectious embolism is not uncommon as it can appear in 20% to 50% of patients and could be asymptomatic and though di cult to recognize (13), (19), (20). In our cohort, PET/CT identi ed cases of clinically unsuspected septic embolism and clinical important extracardiac ndings. It also provided an alternative or additional diagnosis in more than half of the negative cases, including 13 cases of neoplasm previous unknown.

Limitations
This was a single-center study with recognized limitations. Our sample size was insu cient to allow robust subgroup analyses such as the performance of PET/CT in mechanical vs biological prothesis or intracardiac devices vs prothesis. The gold standard for the diagnosis was the clinical judgment of the endocarditis team based on the results of diagnostic tests and the clinical follow-up of patient. Finally, in our study the time between the beginning of antibiotherapy and the PET/CT was longer than reported by other studies.

Conclusions
18F-FDG PET/CT proved to be a useful diagnostic tool in patients with both suspected NVE and PVE with good sensitivity and excellent speci city, resulting in a signi cant decrease in the number of cases of possible EI. This signi cant diagnostic improvement, which could have an important impact on IE outcomes, merits further assessment in larger series.