Fungal endocarditis

Article written and reviewed by Dr Luca Moretti

Fungal endocarditis is an infection of the endocardium caused by yeasts or molds, characterized by diagnostic difficulty, frequent association with prosthetic material and the possibility of late relapse. Candida and Aspergillus are the main groups, but they differ in routes of spread, blood-culture yield, resistance and treatment. The category therefore does not correspond to a single therapeutic disease. A bulky vegetation in a predisposed patient should raise the possibility, but neither size nor absence of bacterial growth alone identifies a fungal cause.

Suspicion increases after cardiac surgery, in the presence of prostheses or devices, candidemia, prolonged vascular access, parenteral nutrition, immunosuppression and injection drug use. Infection can nevertheless involve a native valve without profound immunodeficiency, particularly when an intravascular inoculum or previous valvular damage exists. Prognosis depends on the organism, extent, dissemination and ability to achieve source control. Mortality percentages from small series cannot be directly transferred between Candida and molds, or between operable patients and those with advanced disseminated infection.

Etiology, pathogenesis and pathophysiology

Candida may enter the bloodstream through catheters, damaged mucosa, abdominal foci or other sources and adhere to endothelium and biomaterials. C. albicans remains important, but non-albicans species have substantial clinical relevance. C. parapsilosis has a particular association with devices and healthcare-associated transmission; C. glabrata, C. krusei and other species may have very different susceptibility profiles. Taxonomic nomenclature may vary among laboratories, but identification must permit precise drug selection and comparison with any previous isolates.

Candida biofilm contains cells with different organization and metabolism embedded in an extracellular matrix. The activity of some antifungals against planktonic cells does not fully predict activity against adherent populations; conventional susceptibility of the isolate therefore does not prove that a drug alone will sterilize a prosthesis. In species capable of producing them, hyphae and pseudohyphae contribute to adhesion and invasion. Persistence on foreign material explains both the need for source control and, in selected cases, the usefulness of suppressive therapy after the initial phase.

Aspergillus may disseminate from an invasive focus, most often respiratory, or be associated with contamination during cardiac procedures. Tissue growth and angioinvasion promote necrosis, vascular lesions and emboli. Routine blood cultures are frequently negative despite extensive infection, so valve or embolic tissue may provide the main opportunity for identification. Absence of neutropenia does not exclude endocarditis, especially after surgery or on prosthetic material; conversely, Aspergillus isolated from a respiratory site does not automatically prove cardiac localization.

Less common yeasts and molds, including Scedosporium and other opportunistic pathogens, require specialist identification because they may have intrinsic or acquired resistance that makes a regimen selected for Candida inadequate. Relevant differences also exist within the same genus. Morphological diagnosis of hyphae in tissue supports invasive infection but does not always reliably distinguish species; culture and molecular methods complete the information. When an isolate is rare or the susceptibility profile is unexpected, confirmation in a reference laboratory may concretely alter treatment.

Antifungal resistance should be distinguished from biofilm tolerance. Alterations in azole targets or increased efflux may reduce azole activity; mutations in FKS genes may compromise echinocandins. Expected profiles already differ among some Candida species, but species identification does not replace susceptibility testing of the isolate, particularly after prolonged exposure. Infection developing during prophylaxis may select organisms or strains not covered: continuing the same class without identification risks leaving the focus active. In molds, differences between species and species complexes make confirmation important when response does not match expectations.

Fungal vegetations may become large and friable, combining thrombotic material with microbial growth. Detachment can occlude large-caliber arteries and seed peripheral organs; size should not, however, become a prerequisite for suspicion because antibiotics, embolization or prosthetic location may make the lesion less apparent. Destruction of leaflets, chordae or the prosthetic interface produces regurgitation, while large masses may obstruct opening and cause stenosis. Perivalvular extension forms abscesses and fistulas and may compromise prosthetic stability.

Dissemination may involve the brain, eye, spleen, kidneys, bones and joints. The site depends on the portal of entry, cardiac location and organism; a cerebral lesion may result from an infected embolus, vascular invasion or abscess. Host immune dysfunction contributes to progression, but antifungal availability in different tissues is also decisive: controlling fungemia with a drug does not guarantee effective concentrations in the vitreous or central nervous system. Every documented site must therefore be incorporated into the therapeutic strategy.

Clinical manifestations

Persistent or recurrent fever, chills, asthenia and weight loss may precede heart failure or embolic events. In a patient with candidemia, persistent positive cultures despite therapy and removal of the suspected source requires investigation for endocarditis, thrombophlebitis and deep foci. A single negative follow-up culture does not prove that cardiac localization has been excluded. A new murmur, dysfunctional prosthesis or embolic phenomena increase suspicion; absence of fever in an immunosuppressed patient is not sufficiently reassuring.

Major arterial emboli may be the initial manifestation, with stroke, acute limb ischemia, abdominal pain or coronary involvement. Focal deficits, altered consciousness and new headache require urgent brain imaging and assessment for hemorrhage or infectious aneurysm. In right-sided disease, septic pulmonary emboli, nodules and cavitations predominate. The radiological picture alone cannot distinguish fungal from bacterial infection, and the presence of an Aspergillus lung lesion does not automatically clarify whether dissemination started in the heart or the lung.

Heart failure may develop rapidly because of valvular destruction or after a subacute phase of increasing regurgitation. In prosthetic-valve disease, dehiscence, paravalvular leak and abscess may cause instability even without a dominant intracavitary mass. Ocular symptoms, vertebral or joint pain and skin lesions may indicate dissemination and require further investigation. These sites are not ancillary to the heart: they may alter the antifungal drug, duration, indications for procedures and the possibility of achieving durable control.

Investigations and diagnosis

Multiple sets of blood cultures are collected before therapy whenever possible, and the suspicion is explicitly communicated to the laboratory. Candida is commonly recovered by routine systems, but sensitivity is incomplete; some situations require agreed additional methods. Candida in blood should not be considered simple contamination and requires investigation of the source and compatible metastatic sites. For Aspergillus, negative cultures are common and should not delay use of other evidence. An unexpected isolated mold culture, however, requires confirmation and interpretation because environmental contamination and invasive disease have very different implications.

Transesophageal echocardiography has a central role in the presence of prosthetic material, high suspicion or an inconclusive transthoracic study. Vegetation size and mobility, location, regurgitation, obstruction and perivalvular complications are described. If suspicion remains significant after a negative study, reassessment and, when indicated, complementary imaging are required. Cardiac CT better defines some root lesions and aids surgical planning; PET/CT is particularly useful for prosthetic material and the search for foci. No echocardiographic or metabolic pattern identifies the fungal species with certainty.

Beta-D-glucan is an indicator of invasive fungal infection in appropriate settings, but it does not localize infection to the heart and is not specific for Candida. Sensitivity and false-positive results depend on the organism, population, exposures and method; an isolated result must be interpreted together with clinical probability and repeated when useful. Galactomannan can support the hypothesis of aspergillosis, especially in populations in which the test is validated, but a negative result does not exclude disease and a positive result does not prove valvular localization. Antifungal therapy already started may reduce biomarker yield.

Molecular tests on blood or tissue may identify fungi not recovered in culture, but sensitivity, targets and availability vary. A panel that includes some Candida species is not a universal test for every fungus. On operative material, culture, histology and PCR should be coordinated: some tissue is kept unfixed for microbiological examinations and some for morphological assessment. Appropriate stains may demonstrate yeasts or hyphae and invasion, while sequencing and culture aid identification. A recovered embolic fragment may be equally useful and should not be discarded without considering the required tests.

Antifungal susceptibility testing guides therapy together with species identification and previous exposure. A MIC is interpreted using the appropriate method and criteria; when valid clinical breakpoints are lacking for a rare organism, a susceptibility category should not be forced. Azole or echinocandin resistance and infections developing during prophylaxis may require a change of class and further investigation. Susceptibility of cultured cells does not directly measure the eradication of biofilm, so a favorable result does not eliminate the indication for source control. In relapses, comparing new isolates and susceptibility profiles is useful.

The 2023 Duke-ISCVID criteria include Candida among typical organisms in the context of intracardiac prosthetic material; the number and conditions of positive sets remain those specified by the definition. In Aspergillus infection, absence of a blood-culture criterion may make pathological or molecular evidence from tissue central. Classification criteria should not delay necessary treatment in a patient with strong evidence of invasive infection and a cardiac lesion. The differential diagnosis includes pretreated bacterial endocarditis, sterile vegetations, thrombi and tumors; the mere absence of bacterial growth does not automatically select a fungal diagnosis.

Assessment for dissemination includes neurological examination, brain imaging according to the organism and clinical picture, investigation for abdominal lesions, and osteoarticular studies when indicated. The eye requires particular attention when visual symptoms are present or disseminated candidiasis is suspected; recommendations on systematic screening in candidemia differ among societies, so one approach should not be presented as universally agreed. In sedated patients or those unable to report symptoms, clinical assessment has further limitations. Ocular or cerebral involvement changes the regimen because antifungal penetration is not uniform.

Treatment and prognosis

Treatment combines systemic antifungal therapy, early cardiac-surgical assessment and control of extracardiac foci. Initial choice considers the probable yeast or mold, instability, previous therapy, organ function and involved sites; identification should allow the regimen to be narrowed or corrected. Evidence specific to fungal endocarditis derives largely from cohorts and case series, while many recommendations are strong because of disease severity and clinical plausibility despite limited certainty of evidence. This distinction should be maintained when alternatives are discussed or an inoperable patient is assessed.

In Candida endocarditis, reference initial options are a lipid formulation of amphotericin B, usually 3-5 mg/kg/day, with or without flucytosine, or a high-dose echinocandin. In adults, IDSA regimens include caspofungin 150 mg/day, micafungin 150 mg/day or anidulafungin 200 mg/day; for flucytosine, when used, the reference is 25 mg/kg every six hours, adjusted particularly for renal function and exposure. These doses are not interchangeable with those used for uncomplicated candidemia and require specialist prescribing and monitoring.

The choice between an echinocandin and lipid amphotericin considers susceptibility, toxicity and dissemination. Echinocandins have useful activity against Candida and biofilm but limited penetration into some ocular and neurological sites; involvement of these sites may require a different regimen. Amphotericin may cause nephrotoxicity and potassium and magnesium losses, although lipid formulations have a different profile from deoxycholate. Flucytosine may accumulate in renal failure and cause cytopenias and hepatotoxicity. Adding a drug should have an explicit objective rather than being justified solely by perceived severity.

Fluconazole is not the usual initial monotherapy for active unstable endocarditis. It may be used as step-down consolidation therapy, commonly 400-800 mg/day in adults, when the isolate is susceptible, the patient is stable and candidemia has been controlled. The dose requires renal adjustment and interaction review. If the strain is not susceptible to fluconazole but is susceptible to another azole, appropriate alternatives may be considered with monitoring. Availability of an oral formulation does not by itself demonstrate adequacy of treatment for vegetations or residual foci.

Surgery for valvular candidiasis is generally recommended when feasible, together with prolonged therapy. After surgery, the IDSA reference is at least six weeks of antifungal treatment, with additional duration for abscesses or other complications: this rule should not be confused with the counting used in common bacterial endocarditis. In prosthetic-valve infection, chronic azole suppression should be considered to prevent relapse when susceptibility permits. When replacement is not possible, long-term suppression may maintain control but is not equivalent to proof that infected material has been eradicated.

The 2025 ECMM-ISHAM-ASM global guideline confirms liposomal amphotericin B, with optional flucytosine, or echinocandins as initial strategies and recommends surgical evaluation capable of leading to intervention within the first week after diagnosis, or sooner if necessary. The document distinguishes strength of options and prosthetic or device-associated contexts; the IDSA doses reported above should not be presented as the only regimen shared by all documents. Postoperative therapy of at least six weeks remains central, longer when complications are present, with individualized consolidation or suppression.

The Arnold cohorts and the ESCAPE study illustrate treatment variability and the importance of long-term outcomes. Failure of an observational comparison to show a uniform benefit from surgery or a particular drug does not prove equivalence: small numbers, patient selection, severity and time to intervention may alter the result. Likewise, success in individual cases treated without surgery does not justify generalizing that strategy. The decision should compare the realistic possibility of medical control with the consequences of retaining an intracardiac focus, not only operative tolerability.

In Aspergillus endocarditis, recommendations support early surgery combined with voriconazole or lipid amphotericin B. A usual adult intravenous voriconazole regimen is 6 mg/kg every twelve hours for the first two doses and then 4 mg/kg every twelve hours, with concentration monitoring and adjustment to clinical conditions. Resistance, intolerance and sites of dissemination may change the choice. After valve replacement, very prolonged therapy, sometimes lifelong in selected cases, may be necessary because of relapse risk; echinocandins are not an equivalent primary monotherapy for this endocarditis.

Azole monitoring includes liver function, interactions and, when appropriate, plasma concentrations. For voriconazole, metabolic variability, absorption, neurotoxicity and visual disturbances make exposure monitoring particularly useful. During long treatments, phototoxicity and other cumulative effects also become relevant. Rifampin and other potent inducers may render an azole regimen ineffective; anticoagulants and immunosuppressants may instead require adjustment because of increased exposure. Medication review should be repeated when drugs are added and discontinued because interactions do not always cease immediately.

For rare or resistant fungi, treatment must be individualized with mycology consultation, using identification and susceptibility to select agents or combinations. New antifungals or drugs active in other mycoses should not be presented as validated endocarditis therapy without specific data. Combinations may be necessary in selected cases, but superiority has not been demonstrated for every association. Even when available antifungal options are limited, the possibility of removing material, draining foci and correcting predisposing factors remains a substantial component of care.

Surgery removes emboligenic masses and infected tissue and corrects regurgitation, obstruction or dehiscence. Invasion of the root or fibrous body requires debridement and complex reconstruction, and simply replacing the prosthesis may be insufficient. An indication for shock or refractory heart failure cannot wait for a predetermined antifungal course. Ischemic stroke, hemorrhage or an infectious aneurysm instead alters the balance and requires neurovascular discussion. The embolic risk of fungal disease makes rapid coordination of cardiac anatomy and cerebral status particularly important.

Persistent fungemia requires systematic review of the entire pathway. Identification and susceptibility, doses actually administered, vascular accesses remaining in place and the possibility of thrombophlebitis, abscesses or colonized devices are reviewed. Cultures are repeated to document clearance, but the same therapy should not simply be prolonged if the patient worsens or develops new lesions. Discordance between apparent susceptibility and failure may result from biofilm, inadequate exposure or a second organism. Step-down to an azole for consolidation therefore requires both a favorable microbiological result and clinical stability with a credible source-control strategy.

Infected vascular accesses and devices should be removed whenever possible, together with drainage or control of collections and foci that sustain fungemia. In Candida infection of a pacemaker or defibrillator, removal of the entire system is recommended; a fungal pocket infection does not automatically follow the short antibiotic duration used for a bacterial pocket infection. For non-removable material, such as some circulatory support devices, prolonged regimens and suppression are considered. An access device required for treatment itself becomes a potential risk: maintenance, surveillance and replacement when indicated must form part of the plan.

Ocular, cerebral and osteoarticular sites may require drugs with different penetration, local treatment or additional procedures, and durations longer than those for the heart alone. Candidemia that clears while pain or focal deficits persist requires reassessment of these compartments. Reducing immunosuppression may aid control but is not always possible in transplant recipients or during cancer therapy; the balance should be agreed with the responsible specialists. In people who inject drugs, treatment of substance-use disorder and harm reduction reduce the risk of new inoculations and facilitate completion of therapy.

For ocular candidiasis, the presence of chorioretinitis, macular involvement or vitritis determines different choices and may require intravitreal or surgical treatment in addition to systemic therapy. An echinocandin effective in the bloodstream does not guarantee control of the vitreous compartment. Similarly, brain abscesses, osteomyelitis and arthritis require a specific penetration strategy and, when indicated, drainage. This complexity should not become disconnected regimens: infectious-disease physicians, cardiologists and specialists for the involved organs should agree on a strategy covering all sites without adding avoidable toxicity. Final duration follows the focus requiring the longest treatment, not merely clearance of blood cultures.

Suppressive therapy aims to prevent regrowth or relapse when risk remains high, especially with prosthetic material, residual hardware or inability to operate. The drug must be active against the isolate and sustainable long term, with monitoring of adherence, toxicity and interactions. Suppression should not be stopped solely because echocardiography improves, nor continued without reassessment of the overall strategy. Recurrent fever, new fungemia or increasing lesions during therapy may indicate resistance, inadequate exposure or an uncontrolled source and require new samples whenever possible.

Prognosis remains poor particularly with shock, perivalvular invasion, neurological dissemination, persistent immunosuppression or failure of source control. Follow-up must continue for a long time because relapses can appear months or years later. Valvular function, status of foci, drug tolerability and functional recovery are documented; falling biomarkers may support response but do not by themselves certify sterilization. The pathway should specify who reassesses the patient and which symptoms require rapid access to care, avoiding confusion between the end of hospitalization and the end of risk.

Complications

Cerebral and visceral emboli may cause extensive ischemia, abscesses and hemorrhage from infected vascular lesions. In Aspergillus, angioinvasion may add direct wall injury to simple embolic occlusion. Intravenous thrombolysis is not a recommended strategy in stroke associated with endocarditis; possible thrombectomy and treatment of aneurysms belong to specialist neurovascular assessment. Recovered material may contribute to etiologic identification. Anticoagulation to prevent detachment of vegetations does not correct the mechanism and may increase hemorrhagic risk.

Cardiac destruction includes perforations, rupture of the subvalvular apparatus, abscesses, fistulas and prosthetic dehiscence. Heart failure, hemolysis, shock and conduction disturbances may result. Masses may also obstruct a prosthesis, causing a rapid rise in gradients; flow-related changes, thrombosis and degeneration must be distinguished with imaging and microbiology. Improvement in inflammatory markers does not repair these lesions, so anatomical surveillance remains necessary during and after antifungal treatment.

Persistent dissemination may compromise vision, neurological function, spinal stability and joints. A site poorly penetrated by the initial regimen may become a reservoir even after cardiac surgery. Follow-up should therefore include response of individual organs, drainage when required and verification of pathogen susceptibility. If new signs appear during therapy, the explanation may be progression, embolization of pre-existing material, inflammatory reaction or new infection: distinguishing them requires samples and imaging, not merely an empirical dose increase.

Antifungal toxicity may limit therapy that must continue for a long time. Renal and electrolyte injury from amphotericin, cytopenias from flucytosine, and azole toxicity or interactions require monitoring proportional to the regimen. Any therapeutic modification must maintain adequate coverage and account for disseminated sites, avoiding unprotected intervals. Late relapse may reflect residual material, resistance or inadequate exposure; obtaining a new isolate and comparing it with the previous one help determine whether to change the drug, reopen surgical assessment, or both.