In an oncological setting, myocardial FDG uptake varies from patient-to-patient and even for serial examinations of the same patient. This variation is considered to be due to nonspecific FDG uptake patterns in which diffuse myocardial activity is higher than liver activity or uptake in the lateral wall and/or ring shaped/circumferential basal uptake. Since this physiological uptake can mask pathological activities, patient preparation methods to inhibit physiological uptake have been proposed [45,46]. The degree of myocardial glucose metabolism varies greatly depending on the patient’s overall metabolic status, while the fasting myocardium uses free fatty acids (FFAs) as the main energy source (90%) [47]. Following dietary carbohydrate intake, myocardial metabolism shifts to glucose, following the Randle cycle [47,48]. The myocardium metabolizes glucose when blood sugar and insulin are elevated and FFAs are decreased. Conversely, as glucose and insulin levels decrease and FFAs increase in the fasting state, FFAs are used as the primary source of myocardial energy. The glucose metabolism in inflammatory cells is regulated by glucose transporter 1 (GLUT1) and GLUT3, unlike in the myocardium, where it is regulated via GLUT 4 and is independent of insulin effects [49,50]. Therefore, a patient preparation method that enables myocardial FFA metabolism, while simultaneously suppressing physiologic glucose uptake, is essential for successful FDG PET cardiac infection/inflammation imaging.

Most previous studies have implemented a high-fat, low-carbohydrate (HFLC) diet prior to a prolonged fast, or prolonged fast alone, while some studies have also used additional intravenous heparin administration. Although many preparation methods have been proposed to suppress physiologic myocardial glucose uptake, most studies included a small number of patients, and the preparation strategies were heterogeneous. Therefore, no clear consensus has been reached on the optimal method [51]. The guidelines of Society of Nuclear Medicine and Molecular Imaging, American Society of Nuclear Cardiology, and Society of Cardiovascular CT (SNMMI/ASNC/SCCT guideline) recommend a preparation regimen that incorporates a fat-rich and low-carbohydrate diet for 12 to 24 hours before scanning, and fasting 12 to 18 hours before and/or the infusion of intravenous heparin approximately 15 minutes before FDG injection [52]. More recently, the Japanese Society of Nuclear Cardiology (JSNC) performed an analysis of previous research and suggested a more detailed method, which involved fasting for 12 to 18 hours, and a low-carbohydrate diet, with a total carbohydrate content of less than 5 g for dinner the day before the scan. In addition, they suggested that patients with diabetes underwent same preparation as those without diabetes, albeit with particular attention to sugar control [53].

Regarding the effect of heparin preadministration on inhibition of physiologic uptake, Osborne et al. [51] reviewed 31 dietary preparation studies and found that the myocardium appropriately suppressed FDG uptake in 87% to 93% of patients who fasted for 4 hours after two HFLC meals. In addition, the same effect was reported when unfractionated heparin was injected 15 minutes before FDG injection after at least one HFLC meal and overnight fasting. Moreover, the authors did not recommend the fasting-only methods, food or drink intake, unrestricted diets, and high-fat supplements within 4 hours prior to scanning [51]. Furthermore, heparin showed an anticoagulant effect at doses above 10 U/kg, and the majority of published studies used unfractionated heparin 50 U/kg [54]. In order to lower the risk of heparin-induced thrombocytopenia (HIT), an uncommon but potentially life-threatening risk, low molecular weight heparin has been suggested; however, low molecular weight heparin undergoes approximately 60% lesser lipolysis than unfractionated heparin [55]. The aforementioned SNMMI/ASNC/SCCT guidelines suggest the use of 15 to 50 U/kg heparin, while the JSNC guideline does not recommend the use of heparin. By way of explanation, the JSNC explained that when fasting for more than 18 hours after a low carbohydrate diet, the inhibition of physiological myocardial uptake by heparin was limited, and the risk of HIT following unfractionated heparin is not negligible. However, only a small number of patients were included in the studies cited in these guidelines or review, and studies may provide varying suggestions owing to different fasting times, dietary methods, and frequency; therefore, one is not absolutely correct. The JSNC performed an analysis of fasting over 18 hours by dividing 82 patients into two groups [53], while Scholtens et al. [56] performed an analysis by dividing 150 people into three groups, and found that heparin additionally induced a significant decrease in myocardial intake during 12 hours of fasting.

Intracellular calcium is known to stimulate glucose uptake, and Gaeta et al. [57] reported a significant decrease in myocardial FDG uptake in mice treated with verapamil prior to FDG injection. However, these findings have not been replicated in humans [58].

Studies to date have failed to establish a single preparation technique that is far superior to all others, and metabolic profiles vary among patients; therefore, each center must optimize its own protocols that adhere to the principles described above.

Hyperglycemia has been shown to impair the inflammatory cell uptake of FDG, as a result of competition with endogenous blood glucose; thus, it is recommended to perform scanning when the patient’s blood glucose is below 200 mg/dL. While most studies perform imaging acquisition 1 hour after FDG injection, Caldarella et al. [59] reported that an improved target-to-back-ground ratio is possible if imaging is delayed by 2 to 3 hours. This finding may have additional value for CIED infections, especially in lead infections with low diagnostic sensitivity. However, a patient series study comparing 1- and 2.5-hour post-injection images in PVE patients reported a false-positive interpretation trend for late images, requiring attention to interpretation [60].

Whole-body (head to feet) FDG PET/CT scanning has an advantage with regard to assessing localization of extracardiac infection, including clinically unpredicted distant foci [40,61]. In addition, treatment can be modified depending on the presence or location of the lesions [40]. Although physiological uptake leads to difficulty in visualizing small intracranial lesions, additional infectious lesions have been found on whole-body PET/CT in 17% of patients [13].

Gated cardiac PET imaging can improve spatial resolution and enhance the detection of small moving lesions with the heartbeat, and comparison with CT angiography (CTA) images is easier [62]. However, gated cardiac PET requires a surplus scan time, and there are currently no publications that have examined the additional value of gated cardiac PET for the diagnostic performance of endocarditis.

Pizzi et al. [63] suggested that a combination of FDG PET with CTA could improve the sensitivity in PVE and CIED infection. In their study, they demonstrated that the sensitivity, specificity, PPV, and NPV were 54.5%, 93.8%, 92%, and 60.9% for mDC; 86.4%, 87.5%, 90.2%, and 82.9% for PET/nonenhanced CT; and 91%, 90.6%, 92.8%, and 88.3% for PET/CTA. They also demonstrated that the use of PET/CTA significantly reduced the proportion of suspicious cases to 8%, from 20% in PET/nonenhanced CT. Furthermore, the high sensitivity of FDG PET for diagnosing infections, combined with the high spatial resolution of cardiac CTA, which delineate structural damage, was able to the nine possible cases in PET/nonenhanced CT to be reclassified into eight rejects and one definite case [6]. Moreover, in a study of adult patients with congenital heart disease and suspected IE and/or CIED infection, FDG PET/CTA enhanced the diagnostic sensitivity from 39.1% to 89% and confirmed the diagnosis in 92% of cases [63].

It has been advised that FDG PET/CT is untrustworthy in the 2 months after surgery [64]. Moreover, the ESC guidelines recommended that it should be used with caution in interpreting FDG PET/CT results in patients <3 months after cardiac surgery, as postoperative inflammatory responses may cause nonspecific FDG uptake; thus, in these cases, radio-labeled leucocyte single photon-emission CT (SPECT)/CT can be considered as an alternative [5]. However, recent studies refuted that scanning for an early after the surgery can be correctly displayed true-negative and the waiting may not resolve potential misidentification problem [14,65,66].

Most studies used visual image analysis, to distinguish diseased states from physiologic uptake. The visual evaluations have been reported to have 74%, 91%, 89%, and 78% sensitivity, specificity, PPV, and NPV for PVE, respectively; and were significantly improved by excluding confounders, as described above (91%, 95%, 95%, and 91%) [14]. The most typical finding in infection is an increased localized uptake in the valve annuli, near the valve leaflets, or in prosthetic materials, where FDG uptake is not observed physiologically. In contrast, a mild homogeneous uptake near mechanical prosthetic valves should be considered as a physiologic uptake. However, a high uptake near prosthetic valves, particularly in cases with clinically high suspicion of infection without other infectious foci, should be considered as possible infection, even if homogeneous.

FDG PET/CT can overlook small vegetations, and endocarditis can be ruled out if there is no FDG uptake in the heart. However, if FDG avid metastatic infection lesions, such as septic pulmonary emboli, fungal aneurysms, or brain abscesses, are found, they may be considered as evidence of endocarditis, even in the absence of cardiac abnormalities [67]. This guidance is especially important when FDG PET/CT is less sensitive at the primary focus, such as in NVE or lead endocarditis, as mentioned earlier.

AC artifacts occur when high-density structures produce beam hardening or scattering artifacts in low-dose CT used for AC, and can result in false uptake. Non-AC PET images are difficult to evaluate quantitatively, but if the lesion uptake is less than two-fold that of the liver uptake, the artifacts should be excluded by comparing the non-AC images with the corresponding AC images to reduce the possibility of false positives (Fig. 3) [25]. In addition, metal artifact reduction algorithms can improve the confidence of AC image analysis in patients with metallic cardiac devices or valves [68,69].

Semiquantitative measurements of the metabolic activity of lesions using SUVs are less subjective and provide a more objective threshold for determining infection. For this purpose, Scholten et al. [70] analyzed studies on the SUV value of PVE. However, intercenter exchanges were not possible due to the lack of standardized protocols between studies; and because the threshold for rejected PVEs (0.5 to 4.9) and definite PVEs (4.2 to 7.4) showed wide ranges. A semiquantitative measure of FDG uptake, the European Association of Nuclear Medicine Research Ltd.-standardized SUV value ratio (affected valve/blood pool) of ≥2.0, was a 100% sensitive and 91% specific interpreter of PVE [14]. It has been proposed that grading of valvular FDG uptake as intense, moderate, mild, or absent is more useful than a simple grading of positive or negative. Similar to the mDC for echocardiography, it is also possible to set up a registry to predict the probability of PVE for each uptake intensity [71].

False-positive findings may occur in areas where surgical adhesives have been applied, which may last for 2 or 3 months [14,45]. False-negative findings may be due to low inflammatory activity at the time of imaging or as a result of prolonged antibiotic treatment [14]. Swart et al. [14] reported that a CRP value less than four-fold the upper normal limit (<40 mg/L) was a remarkable false-negative factor. Although PET is highly sensitive to detecting disease activity, it has lower spatial resolution than CT. To visualize sites with an abnormal uptake by PET, the target structure should have a volume larger than 1 cm³, with significant accumulation of the administered radio-tracer [72]. In addition, PET requires data collection for a relatively longer time compared to CT, and motion from the heart beat and breathing during data acquisition can degrade the spatial resolution of PET. However, the detrimental effect caused by motion cannot be noticeably overcome by physiological gating during data collection. Therefore, the use of PET is not optimal for detecting and characterizing small lesions [73]. Overall, clinicians face considerable challenges in diagnosing and characterizing infectious heart diseases with structural or functional imaging, and further improvement in imaging technology will be needed.