Blood culture-negative endocarditis

Article written and reviewed by Dr Luca Moretti

Blood culture-negative endocarditis is a diagnostic syndrome in which an endocardial infection is not documented by routine blood cultures. The definition describes a limitation of the microbiological pathway, not a single causative organism or a particular valvular morphology. It may involve normally cultivable pathogens suppressed by antibiotics, microorganisms requiring different methods, sampling problems or a low circulating microbial burden. The first task is to reconstruct why cultures are negative; the second is to verify that the lesion is truly infectious, maintaining a differential diagnosis with thrombi, tumors and non-infective endocarditis.

A culture that is still negative after only a few hours is not sufficient to define the syndrome. Adequate samples, knowledge of incubation times and interpretation of clinical suspicion are required. A patient already treated for fever may have staphylococcal or streptococcal endocarditis without further blood-culture growth; by contrast, a patient never exposed to antibiotics, with a subacute course and specific exposures, has a different probability of intracellular or fastidious pathogens. These two situations require different diagnostic and empirical strategies and should not be collapsed into a universal antimicrobial combination.

Etiology, pathogenesis and pathophysiology

Previous antibiotic exposure is a common cause of culture negativity. Even a few doses can reduce bacteremia without sterilizing vegetations or prosthetic material, particularly when the regimen is active but inadequate in dose, duration or source control. The history should include outpatient prescriptions, treatments received in the emergency department and medications taken without initial documentation. Recording only “recent antibiotic” is insufficient: drug, route, dose and timing in relation to sampling help determine which organisms may have been suppressed and how subsequent results should be interpreted.

Insufficient blood volume, a limited number of sets, inadequate transport and samples obtained only through an access device can reduce yield or reliability. Modern automated systems recover many organisms once considered difficult, including the HACEK group under usual conditions; it is incorrect to attribute all negative cases to these organisms or to indiscriminately prolong every incubation. Some organisms, including Cutibacterium, may require specific attention, particularly with prosthetic material. Communication with the laboratory helps distinguish a pre-analytical limitation from a biological requirement of the pathogen.

Coxiella burnetii may cause persistent infection localized to the valves, often on a pre-existing substrate. Contact with livestock and contaminated environments increases suspicion, but absence of a recalled exposure does not exclude disease. The acute phase may have gone unnoticed; subsequent endocarditis may produce small or inapparent vegetations and a prolonged systemic syndrome. Vascular infection may coexist or represent another persistent site. A high serological titer must therefore be linked to clinical and anatomical evidence rather than interpreted as proof of valvular localization by definition.

Bartonella henselae is associated mainly with exposure to cats and fleas, whereas Bartonella quintana is linked to body lice and conditions favoring their spread. Epidemiological associations guide the history but do not replace testing. Endocarditic forms may cause valvular destruction and a marked immune response, with glomerulonephritis and autoantibodies mimicking vasculitis. Negative blood cultures and ANCA positivity should therefore not automatically lead to immunosuppression: infection can generate an apparently autoimmune picture and worsen if unrecognized.

Tropheryma whipplei can cause endocarditis without the classic intestinal syndrome of Whipple disease. Long-standing arthralgia, weight loss and constitutional symptoms are useful clues, but fever and diarrhea may be absent. Molecular documentation on valve tissue has an important role; positivity in saliva or stool alone does not prove cardiac infection because it may reflect colonization. Variability of presentation explains why the organism is sometimes identified only after surgery for valvular disease thought to be non-infectious.

Brucella should be investigated in the presence of occupational exposure, animal contact or consumption of unpasteurized dairy products in relevant areas. Nontuberculous mycobacteria, particularly M. chimaera, become relevant after certain cardiac-surgery exposures and may present much later with prosthetic or disseminated infection. These pathways require dedicated cultures and advance communication with the laboratory. Testing should not be extended to every patient with fever: pre-test probability determines yield and the risk of interpreting an ambiguous result as causal.

Among fungi, Aspergillus is an important cause of endocarditis in which routine blood cultures are frequently negative despite invasive lesions and high embolic risk. Candida may instead grow in blood cultures, although sensitivity remains incomplete and results are influenced by therapy. Prosthetic material, cardiac surgery, immunosuppression, catheters and invasive fungal infection modify probability. Not all molds share the same biomarkers or drug susceptibility: a generic diagnosis of “fungal endocarditis” does not replace organism identification when tissue or other samples make this possible.

Microbiological negativity does not attenuate the mechanisms of injury: proliferation within vegetations, valvular destruction, perianular extension and embolization can continue. Some organisms produce a substantial immune response with immune-complex deposition, complement consumption and renal injury; others predominantly invade tissues and vessels. Etiologic delay often prolongs exposure to empirical therapy and may postpone source control. Severity therefore depends on the organism, anatomy and time to effective treatment, not simply on the absence of growth in culture bottles.

Clinical manifestations

The presentation ranges from acute sepsis, often after antibiotics have sterilized the bloodstream, to a prolonged course with low-grade fever, sweats, anemia and weight loss. Endocarditis may be recognized during evaluation of new regurgitation, heart failure or an embolic event. Absence of fever does not exclude infection, particularly in T. whipplei disease, immunosuppressed patients or after treatment. Reconstructing the sequence is useful: preceding arthralgia, development of the murmur, exposures and transient response to medications may be more informative than a single clinical snapshot.

Renal injury with hematuria, proteinuria and reduced complement may reflect immune-complex glomerulonephritis, but also emboli, sepsis or toxicity. Skin lesions, arthralgia and autoantibodies can make the picture resemble rheumatologic disease; interpretation must remain integrated with echocardiography and specialist microbiology. Improvement of inflammation under corticosteroids does not prove an autoimmune origin and may mask infectious progression. When renal biopsy or immunomodulation is required for a severe complication, the decision requires coordination between nephrology and infectious-disease teams.

Cerebral and visceral emboli, spondylodiscitis and abscesses may dominate the course. Their distribution supports dissemination but does not identify the organism and does not always distinguish infected from sterile vegetations. In fungal endocarditis, large-artery occlusions may occur; in prosthetic disease, a new conduction disturbance suggests local invasion even with negative cultures. Progressive dyspnea and instability require immediate definition of the valvular mechanism, without waiting for the etiology to be resolved before discussing necessary surgery.

Investigations and diagnosis

The pathway begins by verifying blood cultures: at least three sets, generally including an aerobic and an anaerobic bottle, with adequate volume and peripheral collection before antibiotics whenever possible. In adults, the volume per bottle follows the system recommendations, commonly about 8-10 mL; underfilling multiple bottles does not compensate for the loss of sensitivity. Previously received drugs and incubation times are reviewed. In an unstable patient, therapy is not withheld to increase yield. A diagnostic antibiotic interruption in a stable, already-treated patient may be considered only in selected circumstances and under specialist supervision, not as a mandatory step.

If after approximately seventy-two hours cultures have not identified the organism and suspicion remains significant, the workup is broadened in agreement with the laboratory; a strongly suggestive exposure may justify targeted testing from the outset. Serology for Coxiella and Bartonella represents a high-yield core in many settings. Brucella, fungi and mycobacteria are investigated according to epidemiology and substrate. Ordering numerous rare tests without a clinical question increases false positives and incidental findings; the strategy should state which hypothesis each test is intended to evaluate and how the result would alter treatment.

In the 2023 Duke-ISCVID criteria, a phase I anti-Coxiella IgG titer greater than 1:800 constitutes major microbiological evidence; for B. henselae or B. quintana, the serological reference is an IgG titer of at least 1:800 by immunofluorescence. Threshold and method must be respected: values obtained with different platforms are not automatically interchangeable. Cross-reactions, antibody persistence and pre-test probability require expert interpretation, with confirmation in a reference laboratory when appropriate. A classification threshold does not make demonstration of disease unnecessary and does not by itself measure activity during follow-up.

Blood PCR for Coxiella, Bartonella or T. whipplei may provide a major criterion under the conditions specified by Duke-ISCVID. A negative result does not exclude valvular localization, particularly with low burden or treatment already started. Sensitivity depends on the organism, matrix, method and timing of sampling. Tests on non-sterile samples require additional caution: detecting DNA in saliva or stool is not equivalent to demonstrating microorganisms within a vegetation. The laboratory should report an interpretable result and avoid turning any molecular signal into an autonomous clinical conclusion.

When available, valve tissue provides an essential diagnostic opportunity. Before surgery, separate samples are arranged for cultures, histology, stains, immunohistochemistry and PCR; some tissue must remain suitable for microbiological testing rather than all being fixed in formalin. Targeted PCR and broad-range amplification with sequencing of bacterial 16S or fungal targets can identify organisms after antibiotics. Tissue yield may exceed that of blood, but contamination and persistence of DNA after cure require comparison with pathology and the clinical picture. The finding should not be interpreted as an antimicrobial susceptibility result when the method does not provide susceptibility information.

Histology may demonstrate active inflammation, necrosis and microorganisms or a predominantly organized thrombotic deposit. Sensitivity of stains varies with organism and treatment; absence of organisms in one section does not prove sterility of the entire valve. Granulomatous findings and macrophages with particular features may suggest specific organisms but require confirmation when possible. In pretreated disease, the combination of morphology and molecular testing may be more informative than culture alone. An unexpected molecular result should also be discussed with the pathologist: a contaminant sequence without a coherent lesion carries a different weight from DNA concordant with demonstrated tissue infection.

Metagenomic sequencing, including analysis of microbial cell-free DNA in plasma, may be useful in selected unresolved cases. Potential advantages are broad target coverage and the possibility of identifying unsuspected pathogens; limitations include variable sensitivity, contamination, non-causal results, cost and availability. The test does not replace properly obtained cultures, appropriate serology or tissue examination. Its clinical value depends on concordance between the detected organism, a plausible site and the patient’s history, and on the possibility of confirmation with an independent method.

Echocardiography should define vegetations, new regurgitation, perforations and complications, with transesophageal examination when indicated and repetition if suspicion persists. Cardiac CT and PET/CT are particularly useful in prosthetic disease and perivalvular extension; positivity does not by itself identify the pathogen. Extracardiac imaging searches for emboli, abscesses, spondylodiscitis and possible portals of entry. A negative PET does not exclude native-valve disease, while prosthetic uptake after surgery requires distinction from sterile inflammation. Microbiology and anatomy answer complementary questions.

2023 Duke-ISCVID clinical combinations:

• Definite endocarditis: two major criteria, or one major and three minor criteria, or five minor criteria.
• Possible endocarditis: one major and one minor criterion, or three minor criteria, in the absence of requirements for definite or rejected disease.
• Specific serological or molecular documentation may provide a major criterion even with negative blood cultures; histology and tissue microbiology may permit a pathological diagnosis.

The prospective Fournier series supported an integrated pathway in which serology, microbiology and valve analysis complement one another. The yield of a test, however, depends on the population referred to the reference center and on tissue availability: it cannot be transferred directly to all patients with fever and equivocal echocardiography. It is useful to distinguish a result that identifies a plausible organism from one that actually provides sufficient evidence of endocarditis. The same principle applies to validation of criteria: greater sensitivity may be accompanied by more cases classified as possible, without all requiring identical treatment.

The differential diagnosis includes cancer-associated NBTE, Libman-Sacks endocarditis, thrombi, fibroelastomas, Lambl excrescences and degenerative lesions. Cancer or lupus does not automatically make a mass sterile, and failure to meet the definite category does not prove NBTE. What matters is convergence among the clinical course, tissue destruction, microbiological evidence and systemic context. If the infectious hypothesis loses consistency after an adequate workup, medications and investigations are reconsidered; continuing antibiotics for weeks solely because the diagnosis remains uncertain may cause harm without addressing the true cause.

Treatment and prognosis

Empirical therapy is constructed by distinguishing pretreated cases from culture-negative cases without prior antibiotics. In the former, staphylococci, streptococci and enterococci remain central, with adjustments for early prosthetic infection, healthcare acquisition and local resistance. In the latter, exposures and phenotype may justify targeted coverage of fastidious pathogens while testing is completed. Recommendations do not propose one regimen applicable to every scenario. Broad antibiotic therapy may be necessary in sepsis, but it should be reassessed as soon as identification, an alternative diagnosis or a concrete gap in coverage emerges.

Coxiella burnetii endocarditis is treated with doxycycline and hydroxychloroquine; the latter modifies the intracellular environment and potentiates the activity of the combination. Usual adult regimens include doxycycline 100 mg every twelve hours and hydroxychloroquine 200 mg every eight hours, with adjustments and specialist monitoring. Treatment is prolonged, generally at least eighteen months for native-valve disease and twenty-four months for prosthetic-valve disease. Clinical response, phase I titers, tolerability, retinal risk and interactions are monitored; persistence of high antibody titers alone does not prove ongoing infection, and a single threshold should not replace assessment of the overall trend.

In persistent Q fever, serological follow-up should whenever possible use the same laboratory and method. A downward trend supports response, but titers may remain high after clinical improvement; conversely, a rising titer together with new symptoms requires investigation for a persistent or vascular focus. Valve replacement alone does not eliminate the need for the specific regimen. During doxycycline therapy, gastroesophageal tolerance, photosensitivity and adherence are addressed; during hydroxychloroquine therapy, retinal toxicity, electrical risk and interactions are monitored. The goal is not to reach an isolated number but to maintain clinical and microbiological control with sustainable drug exposure.

For Bartonella, authoritative documents differ in the proposed strategy. The 2023 ESC guideline describes doxycycline for four weeks combined with gentamicin for the first two; the 2025 AHA statement reports doxycycline, preferred, or azithromycin for twelve weeks together with rifampin for six weeks. The choice requires a coherent regimen and consideration of renal risk, particularly when glomerulonephritis coexists. It is incorrect to construct a protocol by taking the shortest duration from one document and the drugs from another. Valvular destruction, emboli and microbiological response may require surgery regardless of the chosen regimen.

Tropheryma whipplei endocarditis requires prolonged treatment and assessment for possible neurological involvement. Recommendations are not uniform: the 2025 AHA statement describes a four-week intravenous phase with penicillin G or ceftriaxone followed by oral maintenance with trimethoprim-sulfamethoxazole for at least eleven months, whereas ESC includes doxycycline with hydroxychloroquine for at least eighteen months. The choice considers penetration into involved sites, toxicity and relapse risk. Studies of systemic Whipple disease are not automatically equivalent evidence for isolated endocarditis; management requires specific expertise and surveillance after apparent resolution.

For Brucella, prolonged combinations including doxycycline and rifampin with additional agents according to recommendations and the individual case are used, often integrated with surgery when substantial valvular damage is present. Nontuberculous mycobacteria require species identification, susceptibility testing and dedicated combinations for very long durations, together with assessment of infected material. These regimens should not be reduced to a generic extension of empirical antibiotic therapy. Suspicion must also be communicated to the laboratory because of sample-processing requirements and protection of personnel.

Aspergillus and other fungal forms require specific antifungals and source control, as described on the page on fungal endocarditis. Vancomycin, cephalosporins and carbapenems do not become active against fungi by prolonging administration. Positive biomarkers or tissue evidence should therefore prompt timely etiologic reassessment. Species, susceptibility and sites of dissemination determine the drug; an agent active in fungemia may have insufficient penetration into the eye or central nervous system.

Indications for surgery remain heart failure, uncontrolled infection and prevention of embolism under appropriate conditions. Negative cultures are not a reason to wait for etiologic certainty in the presence of acute regurgitation with shock, abscess or fistula. Surgery can simultaneously correct damage and provide diagnostic tissue, but it should not be performed solely to obtain a sample when there is no proportionate clinical indication. The team should plan specimen collection before the operation so that an irreplaceable diagnostic opportunity is not lost through improper preservation.

When the organism is not identified despite an adequate pathway, treatment remains reasoned empirical therapy, with duration and spectrum based on substrate, clinical course and reference documents. It should be documented which pathogens are covered, which have been reasonably excluded and which remain possible. Clinical response to a broad combination does not retrospectively identify the organism, because several pathogens may be susceptible and some symptoms may improve because of other treatments. If response is lacking, reassessment includes alternative diagnoses, drug exposure, collections, resistance and mechanical damage; progressively adding drugs without revisiting these hypotheses may increase toxicity and make diagnosis even more difficult.

Monitoring varies with the organism. When cultures have always been negative, persistent negativity cannot be used as the main proof of efficacy: clinical course, inflammation, valvular function, control of foci and interpretable specific tests are required. Serology may decline slowly; DNA may persist; a vegetation may organize without disappearing. None of these isolated findings automatically justifies stopping or indefinitely prolonging therapy. The plan should specify goals, reassessment intervals and reasons for changing strategy.

Prognosis depends mainly on diagnostic delay, organism, destruction, emboli and the possibility of source control. The label culture-negative endocarditis groups together very different diseases and does not permit a single risk estimate. After the acute phase, months of therapy and years of surveillance may be required, with attention to adherence, interactions and cumulative toxicity. The patient should have a clear point of care for recurrent fever, new neurological symptoms or dyspnea, avoiding undocumented empirical courses that again make the microbiology uninterpretable.

Complications

Valvular destruction may progress during therapy that covers only part of the possible spectrum. New regurgitation, pulmonary edema or a conduction disturbance requires urgent reassessment for perforation, abscess and perianular invasion. Reduction of fever under antibiotics does not exclude these lesions and does not restore valvular competence. The ability to link anatomical evolution with etiologic probability is particularly important when no culture is available to immediately signal inadequacy of the regimen.

Embolic complications include stroke, hemorrhage, infectious aneurysms and visceral infarcts or abscesses. Embolic material recovered during a procedure may provide an additional specimen for histology and microbiological testing. Investigation of the organism should not delay treatment of ischemia or hemorrhage; at the same time, a diagnosis of ordinary thromboembolism should not lead clinicians to ignore a septic origin when the context suggests it. Antithrombotic management and timing of cardiac surgery therefore require multidisciplinary balancing.

Infection-associated glomerulonephritis may be mistaken for primary vasculitis and lead to hazardous immunosuppressive treatment. The problem is particularly relevant in Bartonella disease but is part of the broader spectrum of immunological manifestations of endocarditis. Biopsy, complement, autoantibodies and etiologic tests must be interpreted together. Aminoglycosides, sepsis and contrast exposure may add to renal injury: identifying reversible components helps avoid both stopping a necessary treatment without an alternative and continuing a drug that has become disproportionately toxic.

Late relapses may result from treatment that is too short, a residual focus or inadequate drug exposure. Prevention also requires monitoring for retinal, hepatic and hematologic toxicity and for interactions, according to the regimen used. An adverse event that interrupts a many-month course may compromise eradication as much as an incorrect initial choice. Follow-up must therefore maintain both disease surveillance and pharmacological sustainability, distinguishing a new exposure from persistence of the previous episode.