Introduction
The diagnosis of infectious endocarditis is quite often complex and may be among the most challenging diagnoses facing today’s medical providers. Patients can present with a multitude of clinical signs and symptoms and existing diagnostic testing and criteria are imperfect. Although early diagnosis and intervention are clearly associated with improved outcomes, in nearly 25% of endocarditis cases the diagnosis is made >1 month after onset of symptoms [ ]. Despite the widespread use of tools such as the Modified Duke Criteria, transesophageal echocardiography (TEE), and the newer modality of positron emission tomography (PET), endocarditis remains primarily a clinical diagnosis that is best made when considering a number of variables, including the patient’s risk factors, signs and symptoms, microbiologic data, echocardiographic and radiographic results, and clinical course.
Previously, endocarditis has been classified as having either an acute, fulminant presentation or subacute, indolent course. Currently, clinical presentation is not included as part of accepted diagnostic criteria and generally is not considered with respect to treatment determinations. However, the distinction between these two phenotypes is important for clinicians to consider when approaching a potential diagnosis of infectious endocarditis in that a patient may present as critically ill or may have a several month history of a broad range of signs and symptoms. Additionally, with the increasing number of patients with prosthetic valves, cardiac implantable electronic devices (CIEDs), and left-ventricular assist devices (LVADs), as well as the increased sensitivity of newer diagnostic modalities, such as cardiac PET, providers are often in the position of actively searching for endocarditis before it is clinically apparent. This represents a significant paradigm shift for endocarditis as historically physicians and, in part, the Duke Criteria themselves have relied on exam findings for making diagnoses.
In this chapter we will highlight the clinical features of infectious endocarditis, review existing diagnostic algorithms and testing modalities, and outline a general approach to the diagnosis of this endovascular infection. Perhaps the most important concept we will emphasize in this section is that no one sign, symptom, or test is perfectly sensitive or specific for endocarditis. With this understanding, we would encourage providers to reflect on the entire clinical picture as they approach each patient without overemphasizing specific findings. In this respect, clinicians may find that a multidisciplinary endocarditis team, which approaches this disease from the perspective of cardiac surgeons, cardiologists, infectious diseases, and neurologic specialists, can help in making an accurate diagnosis.
Clinical features
General signs and symptoms
The initial symptoms of infectious endocarditis, regardless of their acuity, are often nonspecific and include fevers, chills, malaise, anorexia, night sweats, dyspnea, headache, and weight loss [ ]. Fever is the most commonly reported symptom and may be present in 90% of patients [ , ]. The degree of fever may vary significantly and cannot be used to rule in or rule out the diagnosis. Notably, patients >60 years of age are more likely to present without fever, which is relevant given that more than half of all endocarditis cases occur in patients over 60 [ ]. The duration of fever, particularly after initiation of appropriate antimicrobial therapy, has been shown to correlate with increased mortality and in 1 series of 26 patients with >2 weeks of fever 27% were found to have an intracardiac abscess [ ]. Dyspnea and cough are frequently reported symptoms and may be associated with congestive heart failure and/or septic pulmonary emboli in cases of right-sided endocarditis.
Musculoskeletal symptoms, including arthralgias (17% of patients) and low back pain (up to one-third of patients), are closely associated with endocarditis and in many cases may precede other symptoms [ , ]. Synovitis both with and without septic arthritis has also been reported in up to 14% of patients [ ]. Additionally the overlapping incidence of spontaneous vertebral osteomyelitis and infectious endocarditis has been reported to be as high as 30.8% [ ]. Consequently, in the absence of an alternative explanation, it is prudent to consider a concomitant diagnosis of endocarditis in patients with vertebral osteomyelitis.
Cardiovascular signs and symptoms
The most common abnormality on physical examination is the presence of a cardiac murmur which can be heard in 85% of patients [ ]. However, this finding is more representative of the patient population who are most commonly diagnosed with endocarditis, many of whom have underlying valvulopathies. In only 8%–15% of cases is a new murmur identified or a worsening murmur detected [ ]. In addition, the proportion of patients with symptoms of heart failure can vary widely. In one series of 40 patients from 1986, 78% of patients were noted to have presented with symptoms of heart failure [ ]. However, in a more recent retrospective data review of 234,731 patients with a primary diagnosis of endocarditis, only 28.9% of patients were found to have a secondary diagnosis of heart failure [ ]. The difference may be accounted for by a variety of changing factors, including the standardization of heart failure definitions and the availability of large electronic databases that allow for data abstraction of a high volume of cases without chart review.
Neurologic signs and symptoms
In addition to cardiac manifestations, neurologic abnormalities secondary to embolic strokes can be found in 10%–35% of left-sided endocarditis patients [ , ]. Given that anywhere from 15% to 70% of all strokes are thought to be embolic and the high morbidity of undiagnosed endocarditis, current American Society of Echocardiography guidelines recommend transthoracic echocardiography or TEE in all patients diagnosed with ischemic stroke [ ]. An area for further study is whether routine blood cultures in the evaluation of stroke could lead to earlier diagnoses of endocarditis. Additionally, approximately 75% of patients will have clinically inapparent neurologic complications of endocarditis including silent emboli, microhemorrhage, mycotic aneurysms, or brain abscess seen on advanced imaging [ ]. Radiographic evaluation for these silent complications will be addressed later in the chapter.
Although not classically considered as a presentation manifestation of endocarditis, meningitis or a meningeal reaction (elevated cerebrospinal fluid white blood cell count with negative cultures) can be seen in 1%–20% of patients [ , ]. It is more commonly associated with Staphylococcus aureus blood stream infection [ , ]. In cases where patients present with clinical signs and symptoms of meningitis it is appropriate to pursue lumbar puncture as the presence of positive cerebrospinal fluid cultures will impact antibiotic selection. Conversely, only approximately 2% of all patients with bacterial meningitis are found to have concurrent endocarditis [ ]. Finally, many patients may present with encephalopathy which may be the result of any of the previously described neurologic phenomenon such as stroke, intracranial hemorrhage, mycotic aneurysm, and brain abscess or could be secondary to a multitude of other factors including illness severity and/or uremia [ ].
Ophthalmologic signs and symptoms
While patients with endocarditis rarely present with primary ocular symptoms, central retinal artery occlusion with associated vision loss is seen as a presenting finding, albeit in <1% of cases [ ]. Endogenous endophthalmitis, which also presents with decreased visual acuity, accounts for 2%–8% of all cases of endophthalmitis. However, nearly 40% of endogenous endophthalmitis cases are caused by infectious endocarditis, and in certain instances this intraocular diagnosis is the primary manifestation of the underlying cardiac infection [ ]. Approximately 3% of endocarditis patients may be subsequently found to have Roth spots, or small white spots, on the retina with associated retinal hemorrhage but these are typically asymptomatic and are only identifiable on dilated eye examination [ ].
Cutaneous signs and symptoms
Dermatologic manifestations of infectious endocarditis have decreased in frequency since the introduction of antibiotics. Classic findings including splinter hemorrhages, or petechial appearing lesions that run parallel to the finger nails, can be identified in as many as 19% of endocarditis cases but this finding is nonspecific as they are also seen in patients with frequent trauma to the hands, mitral stenosis, and renal failure on peritoneal dialysis [ ]. Janeway lesions, or painless macules, seen on the palmar and plantar surfaces are secondary to microembolic phenomenon and can be seen in 2.2% of endocarditis cases. Osler’s nodes, painful erythematous nodules typically on the palms, fingers, or toes, have been hypothesized to be secondary to either immunologic or embolic phenomenon and have fallen from an incidence of 40%–90% in the preantibiotic era to 6.7% in a 1995 retrospective study of 139 endocarditis cases [ , ]. Despite their decreasing prevalence, these dermatologic manifestations retain clinical importance as they are included in the Modified Duke Criteria [ ].
Laboratory evaluation
Blood cultures and microbiology
While there are no serum tests that are perfectly sensitive or specific for infectious endocarditis, the mainstay of laboratory diagnosis are peripheral blood cultures obtained prior to the initiation of antibiotics . Current American Heart Association (AHA) guidelines recommend collection of three sets of blood cultures when evaluating for endocarditis [ ]. The yield of blood cultures increases with the number of samples obtained with literature demonstrating that sensitivity increases from 73% with one culture to 98% with three [ ]. The proportion of endocarditis cases that are blood culture-negative varies widely depending on the study, with some sources estimating that as few as 2% of endocarditis patients have negative blood cultures and others reporting a value closer to 70% [ ]. The 2015 AHA guidelines suggest that approximately 20% of endocarditis cases are culture-negative [ ]. The most common cause for negative blood cultures is their collection after administration of antibiotics. In 2 series comprising a total 204 culture-negative endocarditis cases, 46% of patients received some form of antimicrobial therapy prior to blood culture collection [ , ]. In certain situations it will be clear to the providing physician that blood cultures are indicated. However, in many cases, patients, often with risk factors for endocarditis, will present with the vague nonspecific symptoms discussed previously. As a result, it can be tempting for providers to empirically treat with oral or intravenous antibiotics. We argue that in this patient population, clinicians should have an exceedingly low threshold to obtain blood cultures. While there is an ~1%–5% risk of a false-positive culture, this is outweighed by the significant increases in morbidity and mortality associated with delayed diagnosis of endocarditis [ , ]. This risk of culture contamination is further counterbalanced by the potential side effects and toxicities of broad-spectrum antibiotic administration required in culture-negative endocarditis cases rather than the targeted, narrower spectrum antimicrobial therapy utilized in patients with known pathogens.
With respect to the etiologic causes of endocarditis, gram-positive organisms predominate in several large series of cases, representing >80% of all identified endocarditis pathogens with S. aureus encountered the most frequently [ ]. The changing epidemiology of the disease, particularly with increases in injection drug use–associated cases and widespread utilization of routine antibiotic prophylaxis for at-risk patients, may account for the increasing proportion of endocarditis caused by S. aureus relative to Streptococcal species, which were previously encountered more frequently [ ]. Additionally, a time of <10 h to blood culture positivity has been shown to correlate with a subsequent diagnosis of endocarditis in patients with S. aureus bacteremia. Cultures that turn positive in <13.7 h have also been associated with higher mortality [ ].
In addition to S. aureus , viridans streptococci, other streptococci, such as group A and group B streptococci, nutritionally variant streptococci and Streptococcus gallolyticus , enterococci, and coagulase-negative staphylococci are the other commonly encountered gram-positive pathogens. In a 2011 study of 115 patients with gram-positive bacteremia routine echocardiography leads to a nearly threefold increase in the diagnosis of endocarditis compared to a historical control that underwent echocardiography at physician discretion [ ].
Gram-negative pathogens are a relatively infrequent cause of endocarditis but may account for 1.3%–10% of all cases [ , , , ]. Approximately half of these cases are caused by organisms from the HACEK group ( Haemophilus aphrophilus , Aggregatibacter species, Cardiobacterium hominis , Eikenella corrodens , and Kingella kingae ) with the other half caused by more commonly encountered gram-negative pathogens. Pseudomonas aeruginosa , Escherichia coli , Klebsiella pneumoniae , and Serratia marcescens account for approximately two-thirds of non-HACEK gram-negative cases. Bacteremia with an HACEK organism is reported to have a 60% positive predictive value for a subsequent diagnosis of endocarditis and should prompt providers to evaluate for this possibility. Previously, injection drug use was considered a primary risk factor for non-HACEK gram-negative endocarditis but more recent literature suggests that this population comprises only 4% of such cases and more than half of patients have a health-care exposure as their primary risk factor [ ]. This includes patients with implantable cardiovascular devices such as prosthetic cardiac valves and CIEDs.
Fungal organisms are also a relatively uncommon cause of endocarditis that comprise 2%–4% of all cases with approximately half caused by Candida species [ , ]. The incidence is estimated to be rising due to increasing health-care exposures. Although not included as a “typical” cause of endocarditis, a 2016 series of 187 patients with candidemia found that 5.9% of patients who underwent echocardiography were found to have valvular vegetations [ ].
Additional laboratory evaluation
Complete blood count
As well as blood cultures, additional serum studies, while nonspecific, can be suggestive of a diagnosis of endocarditis. Anemia, typically normochromic and normocytic, may be seen in up to 80% of endocarditis patients [ ]. However, hemolytic anemia can also be seen in cases of prosthetic valve endocarditis due to mechanical destruction of red blood cells secondary to turbulent flow across a diseased valve. Leukocytosis, with a white blood cell count >10,000, is identified in over half of patients with leukopenia seen in 5%–15% of patients [ ]. Thrombocytopenia has also been reported in 5%–15% of patients [ ].
Serum chemistry and urinalysis
Acute kidney injury is another frequent complication of endocarditis that can be detected with routine serum chemistry testing. In one series of 185 patients with endocarditis, one-third developed acute renal failure defined as a serum creatinine >2.0 mg/dL [ ]. Another series of 112 patients reported a rate of acute kidney injury of 68.8% [ ]. Renal failure associated with endocarditis can be due to multiple factors including acute tubular necrosis, septic emboli, drug-induced nephrotoxicity with or without interstitial nephritis, and glomerulonephritis, which can be seen in as many as 22% of patients with endocarditis [ ]. S. aureus endocarditis has been identified as a risk factor for both acute kidney injury and glomerulonephritis [ , ]. In patients with endocarditis-associated glomerulonephritis, 97% present with hematuria and 6% will have nephrotic-range proteinuria [ ]. Hypocomplementemia is commonly identified in this population as 37% will present with low C3 and 16% with both low C3 and C4. Additionally, 28% of patients will have a positive antineutrophil cytoplasmic antibody [ ]. If the urinalysis is suggestive of possible urinary tract infection, obtaining a urine culture may be helpful in identifying a source of the endocarditis. However, these must be interpreted with caution in patients without symptoms who cannot give a reliable history as positive urine cultures may be secondary to high-grade bacteremia rather than the cause of it or may simply be the result of asymptomatic bacteriuria.
Immunologic testing can also be abnormal in cases of infectious endocarditis but is not specific for the diagnosis. In a French series of 56 patients with definitive endocarditis, 56% had elevated erythrocyte sedimentation rate, 84% had elevated c-reactive proteins, and 36% had elevated rheumatoid factors [ ]. In another series of 85 definitive endocarditis patients, a c-reactive protein >40 mg/L was identified as a predictor of mortality as well as major adverse events including acute kidney injury, septic cerebral and noncerebral embolic events, and cardiac complications [ ].
Culture-negative endocarditis testing
As previously discussed, patients with negative blood cultures represent a significant proportion of infectious endocarditis cases. Blood cultures may be sterilized due to antibiotic administration prior to their collection. However, there are a number of microorganisms that cannot be grown on routine blood culture media that are known to cause infectious endocarditis. Bartonella species, including henselae and quintana , and Coxiella burnetii may cause ~0.8% of endocarditis but can only be reliably identified by serologic or PCR testing [ ]. Brucella , Legionella species, and Mycoplasma pneumoniae are also organisms that cannot be reliably identified on standard blood cultures and require serologic or PCR testing from resected valve specimens. Additionally, nontuberculous mycobacteria (NTM) are a rare case of infectious endocarditis that require special acid-fast blood culture media for growth. Although the overall incidence of NTM endocarditis is very low, since 2011 there has been an international epidemic of Mycobacterium chimaera prosthetic valve endocarditis associated with the use of contaminated heater-cooler devices during open cardiac surgery [ ]. Providers should suspect this possible diagnosis in patients with culture-negative endocarditis and a history of previous cardiac surgery after 2011. Finally, individuals may also develop nonbacterial thrombotic endocarditis, a phenomenon characterized by the presence of vegetations on cardiac valves comprised of fibrin and platelet aggregates and devoid of inflammation or bacteria. This diagnosis is most frequently encountered in patients with rheumatologic conditions such as systemic lupus erythematosus or ANCA vasculitis as well as with metastatic malignancies.
Although there are a number of infectious causes of culture-negative endocarditis, it can be difficult for providers to develop a systematic approach to the evaluation of this patient population. A tertiary care center in Marseille created a standardized algorithm for evaluating culture-negative endocarditis that was published in 2017 [ ]. Utilizing serologic testing for Coxiella , Bartonella species, Brucella , Legionella , and Mycoplasma as well as serum broad-range and specific PCR testing and valve broad-range 16 S ribosomal RNA PCR testing, the authors were able to identify an infectious etiology in 138/177 (78%) of patients with definite culture-negative endocarditis.
As part of the routine evaluation of patients with suspected endocarditis, it is reasonable to pursue serum testing including complete blood counts, creatinine, erythrocyte sedimentation rate, c-reactive protein, and complement levels as well as urinalysis. In patients with culture-negative endocarditis serum testing for Bartonella, Brucella, Coxiella, Legionella , and Mycoplasma may help yield a diagnosis. In patients with prosthetic valve culture-negative endocarditis with a history of open cardiac surgery after 2011, acid-fast blood cultures are indicated to rule out M. chimaera infection. In surgically managed culture-negative definite endocarditis cases, a broad range of 16 S ribosomal RNA PCR testing performed on resected valve tissue has a sensitivity of approximately 80% with a false-positive rate of 3% [ ]. The role of 16 S ribosomal RNA PCR testing from serum specimens in the diagnosis of endocarditis is currently unclear.
Echocardiography
Echocardiography, both transthoracic (TTE) and transesophageal, plays a critical role in the diagnosis of endocarditis. It represents the best imaging modality for identifying the valvular vegetations which are the hallmark of the disease. Additional echocardiographic findings that are consistent with endocarditis include paravalvular abscess, prosthetic valve dehiscence, or new valvular regurgitation [ ]. While TEE is considered the “gold-standard” and a highly valuable diagnostic imaging test, it remains imperfect [ ]. TEE is limited by its relative insensitivity of diagnosis in prosthetic valve disease and by its invasive nature and need for patient sedation, which can lead to delay in obtaining the study and may prevent some patients from undergoing the test entirely.
Transthoracic echocardiography
Surface echocardiography is recommended as the first imaging study for the evaluation of possible endocarditis by both the AHA and European Society of Cardiology (ESC) endocarditis guidelines [ , ]. The ESC reports a sensitivity of TTE for the diagnosis of native and prosthetic valve endocarditis as 70% and 50%, respectively. This figure is based primarily on the results of a single 1989 study that enrolled 80 patients with 91 diseased valves [ ]. In this particular cohort, vegetations were identified on 47/69 valves with a reported sensitivity of 68.1%. Notably, TTE only detected vegetations in 6/22 affected prosthetic valves (27.3%). There are a number of other studies from the same time period that report much lower sensitivity for TTE ranging from 28% to 63% [ ]. A more recent 2017 retrospective evaluation of 29 patients with definite S. aureus endocarditis who underwent both TTE and TEE reported the sensitivity of TTE at only 21% [ ]. Due to the posterior location of many paravalvular abscesses, the sensitivity of TTE for this diagnosis is very low, ranging from 4% to 28% [ , ]. As a result, all patients with TTE findings consistent with left-sided endocarditis should undergo TEE to evaluate for annular complications of endocarditis such as abscess, pseudoaneurysm, or fistula formation. The same is true for patients with prosthetic valves and negative TTE for which there is a clinical suspicion of endocarditis. Reports of the high-specificity of TTE are somewhat misleading as echocardiographic evidence of a valvular vegetation is a major component of the diagnostic criteria for endocarditis (discussed later in this chapter) [ ]. The evaluation of specificity is therefore only relevant when the results of TTE are compared to pathologic findings from either surgically resected valves or autopsy. The primary advantages of TTE are that it is widely available, can be obtained quickly, and is noninvasive. Additionally, in patients with right-sided endocarditis TTE has been shown to have similar sensitivity to TEE [ ]. However, if clinicians have a high suspicion for infectious endocarditis it may be prudent to pursue TEE immediately rather than awaiting results of a TTE [ ].
Transesophageal echocardiography
As previously mentioned, TEE is considered the best imaging modality for the diagnosis of infectious endocarditis. However, given the need for patients to be fasting and able to receive sedation, there is significant potential for delays when attempting to obtain TEE images. Depending on the institution, TEE may also not be available on weekends. These potential delays and institutional variations in practice have not been well studied.
At present, the ESC reports a sensitivity for the diagnosis of native and prosthetic valve endocarditis as 96% and 92%, respectively [ ]. When these data are analyzed in more detail there is evidence to suggest that the sensitivity of TEE may be overestimated. The reported high sensitivity is based on four studies conducted between 1989 and 1994 [ ]. All of the studies were conducted prior to the introduction of the Modified Duke diagnostic criteria. In three of the studies, Shapiro et al. Erdel et al., and Shively et al., a very small number of patients underwent surgery or autopsy. As a result, the reported sensitivity in these studies is based on the total number of vegetations detected by echocardiography rather than by pathologic examination, essentially guaranteeing a high-sensitivity rate. The reported TEE sensitivity of 94% by Shively et al. was also based on a cohort of only 16 patients. Mugge et al. was the only study of the four in which all patients underwent surgery or autopsy. In that paper, the reported sensitivity of TEE for native and prosthetic valve endocarditis was 94% and 77%, respectively. Notably, there were only 22 cases of prosthetic valve endocarditis. Subsequent literature has suggested that anywhere from 13% to 44% of all endocarditis cases may present with a negative echocardiogram [ , , ]. Additionally, TEE may be falsely negative early in the course of disease. If clinical suspicion remains high for infectious endocarditis, both consensus guidelines recommend repeating a TEE after 5–7 days [ , ]. While there is literature reporting a very high sensitivity of TEE, clinicians should be aware of the shortcomings of the existing data and not overemphasize TEE findings. This is especially true in light of the recent introduction of cardiac PET as an adjunctive diagnostic imaging modality, particularly in cases of prosthetic valve endocarditis.
Cardiac PET
Cardiac 18-fluorodeoxyglucose PET with computerized tomography (18-FDG PET/CT) is a relatively newer nuclear imaging tool that can assist in the diagnosis of infectious endocarditis. 18-FDG PET/CT uses the dual imaging modalities of PET, which can detect areas of inflammation by identifying inflammatory leukocytes that express a large number of glucose transporters and are very metabolically active, and cardiac CT, which can detect structural abnormalities associated with endocarditis. Several studies have demonstrated that 18-FDG PET/CT can increase the sensitivity of the Duke Criteria for prosthetic valve endocarditis to approximately 90% [ , ]. Retrospective studies have also suggested that in patients with a clinical history suggestive of endocarditis, 18-FDG PET/CT correlates very closely with intraoperative findings of infection [ ].
The utility of 18-FDG PET/CT is considerably lower in patients with native valve endocarditis with a 2020 study of 115 patients reporting a sensitivity of only 22%. When combining PET-CT results with the Duke Criteria for native valves, the reported sensitivity was only 65% [ ]. The authors of this study hypothesized that the increased sensitivity was due to the higher number of polymorphonuclear cells seen on pathologic examination of surgically removed prosthetic valves, whereas fibrotic tissue predominated in native valve specimens. Another drawback of cardiac PET-CT is that it may identify physiologic low-intermediate FDG uptake in prosthetic valves that were surgically placed within 12 months from the time of the imaging study [ ]. However, there is also literature demonstrating that increased FDG uptake postoperatively is more closely related to the use of surgical adhesives rather than the prosthetic valve itself and that there is a characteristic pattern found with noninfectious postoperative uptake [ , ]. When interpreted by experienced providers, 18-FDG PET/CT performed in patients with recently placed prosthetic valves may still have utility in the diagnosis of endocarditis. Currently the ESC includes cardiac 18-FDG PET/CT in its prosthetic valve endocarditis diagnostic criteria, provided that it is obtained 3 or more months after surgical valve replacement [ ].
Despite its benefits, cardiac 18-FDG PET/CT is not widely available at North American institutions. This is in part due to the expensive equipment required to perform the procedure and also due to the small number of providers who can expertly interpret the results. Additionally, the ideal PET study is performed after the patient has undertaken a carbohydrate restricted diet for 24–36 h with a period of fasting prior to the study [ , ]. This required preparation may result in delays in care which can potentially lead to adverse clinical outcomes. While 18-FDG PET/CT may be helpful in the diagnosis of endocarditis, its utility is primarily seen in a select subset of patients and it requires significant expertise to accurately interpret the results. With further study and more widespread utilization it has the potential to benefit a significant number of patients.
Computerized tomography
Although not traditionally considered as part of the diagnosis of infectious endocarditis, multisliced ECG-gated cardiac CT has been shown to perform similarly to TEE with respect to identifying valvular vegetations [ , ]. In certain circumstances it may be superior to TEE for detection of paravalvular abscesses and pseudoaneurysms [ , ]. However, the primary role of CT is with respect to identification of embolic complications of infectious endocarditis. CT can identify embolic events to the central nervous system (CNS), lungs, kidney, and spleen. CT-angiography (CT-A) can also identify peripheral vessel occlusions and mesenteric ischemia which may be caused by septic emboli. When patients have clinical signs and symptoms suggestive of embolic phenomenon, there is little disagreement about the role of targeted CT for evaluating these complications. There is debate about the role of empiric CT studies of the head, chest, abdomen, and/or pelvis to evaluate for emboli in asymptomatic individuals. In a 2015 Egyptian study of 81 patients with definite left-sided endocarditis who underwent routine CT-A, 63% were found to have evidence of CNS embolization [ ]. One-third of these patients were clinically asymptomatic. The investigators reported that the CT-A findings led to a change in the management plan for 25.6% of all enrolled patients. However, it is unclear what percentage of clinically asymptomatic patients had their therapies altered as a result of the CT-A images and the incidence of acute kidney injury was not reported. Theoretically, CT/CT-A could help upstage a subset of patients with Duke Criteria possible endocarditis to definite endocarditis if embolic phenomenon were identified. Perhaps the most important role of CT imaging is the detection of intracranial hemorrhage and/or mycotic aneurysms, the presence of which could substantially alter surgical management. Notably, intracranial hemorrhage can be detected on CT without iodinated contrast, thereby avoiding the risk of contrast induced nephropathy. Currently, there is limited data on the role of empiric CT scans of the chest, abdomen, and/or pelvis. Routine utilization of these studies must be weighed against the risk of kidney injury and radiation exposure.
Magnetic resonance imaging
Compared to CT, magnetic resonance imaging (MRI) is more sensitive for the detection of cerebrovascular complications of endocarditis [ ]. Nearly two-thirds of patients undergoing routine cerebral MRI are found to have embolic lesions [ ]. Approximately 30% of those patients are clinically asymptomatic. Similar to CT, the presence of embolic complications on cerebral MRI can upstage the diagnosis of endocarditis. In one series of 53 patients with nondefinite endocarditis systematic brain MRI led to an upgraded endocarditis diagnosis in 32% of cases. These MRI findings also led to changes in the therapeutic plan for 18% of patients [ ]. Risk factors for CNS embolization include vegetation size greater than 10 mm, S. aureus bacteremia, advanced age, diabetes, and atrial fibrillation [ ].
Abdominal MRI has also been shown to detect silent systemic emboli in approximately 34% of patients, a considerably lower rate than brain MRI [ ]. A 2012 French study evaluating the role of routine cerebral and abdominal MRI in 58 suspected endocarditis cases found that abdominal imaging by itself only led to an upgraded diagnosis in one patient. In no patients was the therapeutic plan modified based solely on the results of the abdominal imaging. While the data are limited, there is evidence to suggest that systematic cerebral MRI in patients with suspected endocarditis may aid clinicians in confirming the diagnosis and lead to alterations in care plans. However, these benefits do not appear to extend to routine abdominal imaging. Currently the AHA and ESC guidelines recommend that all patients with suspected endocarditis and headache, neurologic deficits, or meningeal symptoms undergo cerebrovascular imaging [ , ].
Angiography
In patients with suspected or confirmed endocarditis the evidence of subarachnoid hemorrhage on CT or MRI raises concern for the presence of a mycotic aneurysm. Conventional angiography is generally considered the “gold-standard” for the diagnosis of mycotic aneurysms but carries with it the risk of stroke, intracranial hemorrhage, arterial dissection, and puncture-site hematoma [ ]. Both CT-A and MR-angiography (MR-A) can be utilized to evaluate for mycotic aneurysm without the risk of these complications. A 2018 meta-analysis of 10 studies incorporating 868 patients found that CT-A had higher diagnostic sensitivity and accuracy compared to MR-A [ ]. CT-A may also be more readily available than MR-A depending on this institution. However, MR-A utilizes less contrast and carries lower risk of nephrotoxicity.
Electrocardiography
Although not particularly sensitive, electrocardiography (ECG) can provide significant information in patients with suspected endocarditis. Coronary embolization resulting in myocardial infarction occurs in 2.9%–10.6% of all endocarditis cases and can be detected by ST segment changes [ ]. The presence of pericarditis can also be detected by the presence of ST segment elevations. Endocardial abscess may be present in up to 30% of endocarditis cases and is most commonly associated with the aortic valve [ ]. These abscesses can extend into the cardiac conduction system and lead to atrioventricular block in approximately 25% of patients. ECG has a reported positive predictive value of 88% for abscess in patients with suspected endocarditis but has a relatively low sensitivity of 45% [ ].
Diagnostic criteria
Modified Duke Criteria
The first commonly utilized set of standards for the diagnosis of endocarditis were the Von Reyn Criteria, first published in 1981. These criteria relied primarily on microbiologic and exam findings and were introduced before the widespread adoption of echocardiography. As a result, the Duke criteria were developed in 1994 and updated in 2000 [ , ]. Although they were developed primarily to provide better entry criteria for epidemiologic studies and clinical trials, these Modified Duke Criteria have been used as the main diagnostic criteria for endocarditis since their publication ( Table 2.1 ). The algorithm uses a combination of major microbiologic and echocardiographic findings as well as minor clinical and microbiological criteria to stratify patients as having definite, possible, or rejected endocarditis. Major criterion includes (1) persistently positive or multiple positive blood cultures for an organism typical for endocarditis or (2) evidence of endocardial involvement defined by echocardiographic findings or new valvular regurgitation. Minor criterion includes (1) fever, (2) risk factors for endocarditis, (3) immunologic complications, (4) vascular complications, and (5) positive blood cultures that do not fulfill the major criteria. Cases with two major, one major, and three to four minor or five minor criteria are considered to have definite endocarditis. Those with one major and one to two minor or three to four minor criteria are classified as possible endocarditis. All other cases are considered to have rejected the diagnosis of endocarditis.
