Luschka, in 1852, first suggested that IE resulted when septic coronary emboli lodged in the vessels of the cardiac valve.⁴⁷ This hypothesis was discarded, because cardiac valves are poorly vascularized.^45,48,49 It now is clear that the initial colonization occurs on the damaged endothelial surface of the valve. In experimental animals, it is almost impossible to produce IE with intravenous injections of bacteria, unless the valvular surface first is perturbed. If a polyethylene catheter is passed across the aortic valve of a rat or rabbit, IE is produced with intravenously injected bacteria or fungi.^50,51 Microscopic examination of this early lesion shows the organisms intimately adherent to fibrin-platelet deposits overlying interstitial edema and mild cellular distortion that have formed in areas of valvular trauma.⁵² Scanning electron micrographs of the damaged valvular surface confirm the adhesion of microorganisms to these areas of fibrin-platelet deposition early in the disease course.⁵³ The organisms are covered rapidly by fibrin.⁵⁴

Opossums and pigs are the only animals known to develop IE readily (i.e., without experimentally induced valvular alteration).^46,55 The stress of captivity is apparently sufficient in these animals to produce subtle valvular changes that lead to spontaneous endocarditis and a markedly increased susceptibility to the disease after intravenous injection of bacteria. In other animals and probably in humans, alteration of the valve surface is a prerequisite for bacterial colonization. Angrist and Oka⁴⁸ first recognized the importance of these deposits as the crucial factor in allowing bacterial colonization of valve surfaces and suggested the term NBTE. Many forms of exogenous stress produce these lesions experimentally, including infection, hypersensitivity states, cold exposure, simulated high altitude, high cardiac output states, cardiac lymphatic obstruction, and hormonal manipulations.⁴⁶ These procedures all increase the susceptibility of the animals to IE.

NBTE has been found in patients with malignancy (particularly pancreatic, gastric, or lung carcinoma) and other chronic wasting diseases, rheumatic and congenital heart disease,⁴⁸ uremia, and connective tissue diseases such as systemic lupus erythematosus (SLE) and rheumatoid arthritis; after the placement of intracardiac catheters (e.g., Swan-Ganz); and after a self-limited acute illness. NBTE generally reflects one of two pathogenic mechanisms: hypercoagulability or endothelial damage. In a careful analysis performed in Japan, NBTE was found in 2.4% of 3404 autopsies, especially in elderly people with chronic wasting disease.⁵⁶ Among patients at high risk for NBTE, the rate may be greater. Cardiac valvular vegetations were found in 19% of 200 nonselected ambulatory patients with solid tumors undergoing prospective echocardiographic screening.⁵⁷ Valvular lesions were most common among patients with carcinoma of the pancreas or lung or lymphoma.

NBTE is most common on the low-pressure side of the cardiac valves along the line of closure, precisely the site most often involved in IE. Whether this lesion is always essential for the development of IE in humans is unknown. Secondary hypercoagulable states (e.g., as seen in secondary antiphospholipid syndromes in SLE) may contribute to the development and propagation of the NBTE lesion.⁵⁸

Hemodynamic Factors

When associated with valvular insufficiency, IE characteristically occurs on the atrial surface of the mitral valve and the ventricular surface of the aortic valve. Rodbard⁵⁹ showed that this localization is related to a decrease in lateral pressure (presumably with decreased perfusion of the intima) immediately downstream from the regurgitant flow. Lesions with high degrees of turbulence (e.g., small ventricular septal defect with a jet lesion, valvular stenosis resulting from insufficient valves) readily create conditions that lead to bacterial colonization, whereas defects with a large surface area (large ventricular septal defect), low flow (ostium secundum atrial septal defect), or attenuation of turbulence (chronic congestive heart failure [CHF] with atrial fibrillation) rarely are implicated in IE. Cures of IE achieved with ligation of an arteriovenous fistula or patent ductus arteriosus without further treatment also highlight the importance of hemo­dynamic factors. A hyperdynamic circulation itself, such as that developing after experimentally induced arteriovenous fistulas in dogs or after creation of fistulas and shunts in hemodialysis patients, may lead indirectly to IE by producing NBTE.^45,46 Finally, implantable intra­cardiac prosthetic material may well contribute to a turbulent blood flow state, as seen in intraventricular pacemakers and implantable cardioverter-defibrillators.

The degree of mechanical stress exerted on the valve also affects the location of the IE.⁶⁰ In 1024 autopsy cases of IE reviewed through 1952, the incidence of valvular lesions was as follows: mitral, 86%; aortic, 55%; tricuspid, 19.6%; and pulmonic, 1.1%. This incidence correlates with the pressure resting on the closed valve: 116, 72, 24, and 5 mm Hg, respectively.

Transient Bacteremia

In the setting of preexistent NBTE, transient bacteremia may result in colonization of these lesions and may lead to the development of IE.⁶¹ Transient bacteremia occurs whenever a mucosal surface heavily colonized with bacteria is traumatized, such as occurs with dental extractions and other dental procedures or with gastrointestinal, urologic, or gynecologic procedures (Table 82-1).^61,62 The degree of bacteremia is proportional to the trauma produced by the procedure and to the number of organisms inhabiting the surface; the organisms isolated reflect the resident microbial flora. The bacteremia usually is low grade (≤10 colony-forming units [CFU]/mL) and transient; the bloodstream usually is sterile in less than 15 to 30 minutes.

In two studies in which samples for blood cultures were drawn from patients with severe gingival disease before the dental procedure, spontaneous bacteremia was identified in 9% to 11% of the patients. Other studies have shown an even higher frequency of spontaneous bacteremia. Of the blood cultured from healthy people, 60% to 80% of specimens were positive when filters and anaerobic techniques were used.⁶³ The degree of bacteremia was low, however, with only 2 to 10 CFU/5 mL of blood isolated. “Nonpathogenic” organisms, such as Propionibacterium acnes, Actinomyces viscosus, Staphylococcus epidermidis, and other Actinomyces or streptococcal species, were responsible. Frequent episodes of silent bacteremia also are suggested by the identification of circulating humoral antibodies to the resident oral flora and by the noted increase in sensitized peripheral T cells to the flora of dental plaque. The frequency of such silent bacteremias in the probable pathogenesis of IE has contributed to the current concept that individual dental procedures are uncommonly the cause of such infections.⁶⁴

Another crucial factor during the transient bacteremia stage is susceptibility of the potential pathogen to complement-mediated bactericidal activity. Only “serum-resistant,” gram-negative aerobic bacilli (e.g., Escherichia coli, Pseudomonas aeruginosa, Serratia marcescens) reliably produce experimental IE in rabbits,^65,66 and this property is found in all isolates from human cases of IE. Although experimental IE can be induced in rats with “serum-sensitive” E. coli, the organisms are eliminated rapidly on catheter removal.⁶⁶

Microorganism–Nonbacterial Thrombotic Endocarditis Interaction

The ability of certain organisms to adhere to NBTE lesions is a crucial early step in the development of IE. Gould and associates⁶⁷ showed that organisms commonly associated with IE (enterococci, viridans streptococci, S. aureus, S. epidermidis, P. aeruginosa) adhered more avidly to normal canine aortic leaflets in vitro than did organisms uncommon in IE (Klebsiella pneumoniae, E. coli). In addition, S. aureus and the viridans streptococci produce IE more readily than does E. coli in the rabbit model of IE.⁶⁸ This observation correlates with the relative frequency with which these organisms produce the disease in humans. The rarity of IE due to gram-negative aerobic bacilli also may be related to their serum sensitivity, as noted previously.

Differences in the propensity to cause IE are apparent even within a single species. For example, specific clones of S. aureus, including clonal complexes 30^69,70 and 22⁷¹ have been reported to be associated with an increased risk for IE. In addition, only 2 of the 11 capsular serotypes of S. aureus described to date, serotypes 5 and 8, account for approximately 75% of clinical isolates, whereas highly mucoid strains (e.g., serotypes 1 and 2) are rarely recovered. Nevertheless, mutant strains devoid of microencapsulation had significantly lower median infective dose (ID₅₀) values in a rat (catheter-induced NBTE) IE model,⁷² compared with wild-type parent strains. Microcapsule expression may attenuate S. aureus IE production by blocking bacterial cell surface adhesins, but this hypothesis requires confirmation. In addition, an increasing number of reports suggest that other specific pathogen characteristics in S. aureus are associated with the severity of infection caused by these strains in humans.^73–77

Other studies using an elegant experimental model of IE after dental extraction in rats with periodontitis, which closely resembles the presumed pathogenetic sequence in humans, also suggest an important role for bacterial adhesion to NBTE in the early events. Although viridans streptococci were isolated much more commonly than group G streptococci in blood cultures obtained 1 minute after extraction, the latter strains caused 83% of the IE episodes in this rat model.^78,79 This propensity to cause IE was associated with an increased adhesion of group G streptococci to fibrin-platelet matrices in vitro.⁷⁹

The adherence of oral streptococci to NBTE may depend on the production of a complex extracellular polysaccharide, dextran. This polymer plays an essential role in the pathogenesis of dental caries by Streptococcus mutans.⁸⁰ Dextran allows the organism to adhere tightly to the surface of dental enamel. The enhanced ability to adhere to inert surfaces also may be important in IE. In an analysis of 719 cases of streptococcal infections in the United Kingdom, 317 cases of IE were found.⁸¹ The most common etiologic agents were Streptococcus sanguinis (16.4% of the cases), previously identified in Streptococcus subacute bacterial endocarditis, and S. mutans (14.2%). Ratios of endocarditis to nonendocarditis bacteremia caused by particular organisms have been derived (Table 82-2), allowing prediction of the relative propensity for a particular organism to cause IE. The ratios range from 14.2 : 1 for S. mutans to a reversed ratio of 1 : 32 for S. pyogenes. Only the first four organisms listed in Table 82-2 (all with ratios >3 : 1) produce extracellular dextran. This finding suggests that dextran production may be another virulence factor in the pathogenesis of IE.

The role of dextran in the adherence of oral streptococci to NBTE also has been studied in vitro with the use of artificial fibrin-platelet matrices (simulating NBTE). The amount of dextran produced by organisms grown in broth correlated with adherence; the amount was increased by incubating the organism in sucrose (which stimulates dextran production) and decreased by the addition of dextranase (which removes the dextran from the cell surface). The addition of exogenous dextran to S. sanguinis grown in sucrose-free media increased adherence. Dextran production also correlated directly with the ability of these organisms to produce IE in vivo in the rabbit model.⁸² The strain of S. sanguinis produced IE less readily when incubated in dextranase than did control strains, and a strain that produced large quantities of dextran produced endocarditis more easily than did a strain that produced relatively small quantities of dextran. Dextran production also increased the adherence of S. mutans to traumatized canine aortic valves in vitro,⁸³ an effect that was dependent on polymers of higher molecular weight.⁸⁴ Dextran formation (or, more properly, exopolysaccharide or glycocalyx production) by oral streptococci may be a virulence factor for the production of IE by these organisms.⁸⁵ Continued in vivo synthesis of exopolysaccharide during experimental IE correlated with vegetation size and resistance to antimicrobial therapy.^86,87 Measurement of cell-adherent glycocalyx by a quantitative spectrophotometric tryptophan assay among viridans streptococci isolated from blood cultures has potential value as an independent predictor of the likelihood of IE.⁸⁸ Non–dextran-producing streptococci may produce IE in humans and adhere to artificial fibrin-platelet surfaces in vitro,⁸⁹ suggesting that other microbial surface characteristics are instrumental for this early event. Whatever the role of the extracellular glycocalyx in microbial adhesion, its presence may retard antimicrobial therapy for streptococcal endocarditis (see later discussion).^86,87,90

FimA, a surface adhesin expressed by viridans streptococci, has been shown to mediate the attachment of such organisms to platelet-fibrin matrices in vitro and to experimental NBTE lesions in the animal model of IE.⁹¹ Homologues of the fimA gene are widely distributed among clinical strains of viridans streptococci and enterococci, suggesting its importance in IE.⁹² Several lines of experimental evidence have confirmed further the key role of FimA in the pathogenesis of IE. Inactivation of the fimA gene has yielded viridans streptococcal mutants exhibiting a significant decrease in virulence in experimental IE compared with the parental strain having intact FimA expression.⁹¹ In addition, animals either passively immunized with anti-FimA antibody or actively immunized with a FimA vaccine were significantly less susceptible to experimental IE than nonimmunized controls.⁹²

Recent data have emphasized the unique interaction of the viridans group streptococci with human platelets in the pathogenesis of IE. The Sullam laboratory published a series of elegant experimental studies^93–95 that supported the following paradigms in viridans group streptococcal IE. First, they identified a group of genes in one viridans strepto­coccal species (Streptococcus gordonii, formerly known as S. sanguis I) that mediates the binding of this organism to human platelets. This locus encodes GspB, a large, serine-rich cell wall–anchored glycoprotein; four proteins mediating the glycosylation of GspB; and seven proteins (the accessory Sec system) that mediate the export of GspB to the bacterial surface. GspB is the adhesin that binds platelet glycoprotein Ib, by means of its high-affinity interaction with sialic acid–containing motifs on glycoprotein Ib. GspB is an unusual protein, not only because of its size (286 kDa) but also because glycoproteins are rare in bacteria. Glycosylation is required for the stability of GspB; loss of carbohydrate results in rapid degradation. This modification, however, renders it an unsuitable substrate for export by the canonical general secretory (Sec) pathway. In contrast, the accessory Sec system is a specialized pathway whose sole function is translocation of GspB from the cytoplasm to the bacterial surface, where it is covalently linked to the cell wall by its LPRTG C terminus. Loss of GspB expression was linked with decreased virulence in experimental IE. Similar results have been reported for Hsa, a homologue in another viridans streptococcal species (S. gordonii).

Second, by transposon mutagenesis, the Sullam group discovered that binding of viridans streptococci to mammalian platelets is mediated in part by two phage-encoded proteins, PblA and PblB, that are associated with the cell wall of this organism. These proteins appear to be important for virulence, in that loss of their expression was associated with decreased virulence in an animal model of IE. This was a surprising finding, in that these are two phage proteins of a lysogenic phage (SM1) that resides on the chromosome of the strain (SF100). PblA is a tape measure protein (important for phage morphogenesis), whereas PblB is a tail fiber protein. Both proteins are linked to the bacterial surface through their interactions with choline groups within the cell wall. A puzzling issue was how these phage proteins get out of the bacterium and onto its surface. It appears that there is some constitutive, low-level expression of the phage lytic cycle, which results in the expression of phage holin and lysin. The net effect is to render the bacteria more permeable, so that PblA and PblB leak out of the bacteria and then bind back to the surface, where they can serve as adhesins. This is a remarkable system, because it is perhaps the first example of a phage-encoded bacterial surface structure and adhesin and because it suggests a co-adaptation or coevolution of the bacterium and the phage, wherein the bacterium is essential for phage replication but the phage also gives the bacterium a selective advantage by providing it with an adhesin.

A similar important role of adhesion to NBTE in the pathogenesis of IE has been shown for yeasts. Candida albicans adheres to NBTE in vitro and produces IE in rabbits more readily than does Candida krusei, a nonadherent yeast rarely implicated in IE in humans.⁹⁶ Although microbial adhesion is a crucial early event in the pathogenesis of IE, the precise intracardiac loci are unknown and may differ among organisms. Most organisms probably adhere initially to a constituent of NBTE; some evidence implicates fibronectin as the host receptor within NBTE.⁹⁷ More recent studies^98,99 have supported this concept. Low-fibronectin–binding mutants of S. aureus and S. sanguinis had decreased ability to produce IE in rats, compared with high-fibronectin-binding parent strains. Other normal constituents of damaged endothelium or NBTE (e.g., fibrinogen, laminin, type IV collagen¹⁰⁰) also may serve to bind circulating bacteria. Streptococcus defectivus, the major species isolated in cases of IE caused by nutritionally variant streptococci (discussed later), bound the extracellular matrix of fibroblasts and endothelial cells in a saturable-specific manner, whereas Streptococcus adjacens and serotype III nutritionally variant streptococcal strains did not bind.¹⁰¹ A study also documented binding of S. mutans, Streptococcus mitis, S. sanguinis, and Enterococcus faecalis to this extracellular matrix. Laminin-binding proteins (e.g., a 145-kDa protein found in S. gordonii) were identified on the cell walls of organisms recovered from patients with IE,¹⁰² and the level of protein expression seemed to be regulated by the presence of extracellular matrix proteins.

Other organisms may bind directly to, or become ingested by, endothelial cells as the initial event.^103–106 This sequence appears to be important in the initiation of IE by S. aureus on “normal” cardiac valves or on native endothelial surfaces adjacent to damaged endothelial sites. Many studies in experimental IE, using S. aureus as the study organism, have shed additional light on the importance of microbial binding to specific matrix proteins found within the NBTE lesion on the development of IE. It seems that the key adhesin possessed by the organism for induction of IE is one or more of its several fibrinogen-binding proteins (e.g., clumping factor, coagulase^107,108). Adhesins for other matrix molecules (e.g., fibronectin, collagen, thrombospondin^109–111) are not involved pivotally in initial attachment of the organism to damaged endothelium but are crucial in persistence of the microbe at this site. Additional virulence factors produced by this organism (α-toxin¹¹²) have been identified in the experimental IE model as important for persistence and proliferation of the organism within maturing vegetations in the next stage of infection, after valvular colonization. The fibronectin-binding adhesins of S. aureus seem to be crucial in the ability of the organism to invade cardiac endothelium and induce endothelial apoptosis,^113–115 although the specific microbial surface–host receptor ligand relationship remains incompletely defined for all the major IE pathogens. This is an active area of investigation, because inhibition of these events may provide novel prophylactic strategies.

The importance of adherence characteristics in the development of endocarditis also has been examined through the use of preincubation of organisms with antibiotics. Many classes of drugs, after incubation even at subinhibitory concentrations, decrease the adhesion of streptococcal species to fibrin-platelet matrices and damaged canine valves in vitro.¹¹⁶ Several elegant studies in animal models verified the significance of this in vitro observation: preincubation of the organism in subinhibitory antibiotic concentrations prevented the development of IE in vivo.^117,118 This finding has direct relevance to the chemoprophylactic prevention of IE (see Chapter 85). In one study, subinhibitory concentrations of penicillin were found to result in a loss of streptococcal lipoteichoic acid, with reduced adhesion to NBTE-involved tissue and an impaired ability to produce IE in vivo.¹¹⁹ Antibiotics may prevent IE by at least two mechanisms—bacterial killing and inhibition of adhesion to NBTE-involved tissue.¹²⁰

Because platelets and fibrin are the major constituents of NBTE, the role of the platelet in the pathogenesis of IE also has been studied. Some strains of bacteria have been found to be potent stimulators of platelet aggregation and the release reaction.¹²¹ In general, IE-producing strains of staphylococci and streptococci more actively aggregate platelets than do other bacteria that less frequently produce IE. Bacteria-platelet aggregates have been found in the peripheral blood of patients with bacteremia. The importance of these aggregates in the formation of the vegetation (or, conversely, the effect of the aggregation on the rate of removal of organisms from the circulation) is unknown. In one study, even small numbers of platelets greatly increased the adherence of oral streptococci to fibrin in vitro.⁸¹ Other studies¹²² showed that S. sanguinis, an important cause of IE, aggregates platelets and adheres to these blood components by means of protease-sensitive components, not dextrans. A platelet receptor for ligands on certain strains of S. sanguinis was suggested. However, this platelet aggregation by viridans streptococci requires direct platelet binding and plasma components.¹²³ Other experiments implicated immunoglobulin G (IgG) in this specific streptococcal bacteria-platelet interaction and suggested that platelet activation is mediated through the platelet surface Fc receptor, with a molecular weight of 40,000 daltons.¹²⁴

After colonization of the valve occurs and a critical mass of adherent bacteria develops, the vegetation enlarges by further platelet-fibrin deposition and continued bacterial proliferation. There is a complex interplay among factors responsible for bacteria-platelet adhesion and aggregation. The ability of S. sanguinis to induce platelet aggregation in vitro is conferred by two bacterial cell surface antigens: (1) class I antigen, which promotes adhesion of S. sanguinis to platelets (adh⁺), and (2) coexpression of class II antigen, which promotes platelet adhesion or platelet aggregation (agg⁺). At least nine adh/agg phenotypes have been identified among naturally occurring variants, reflecting a range of platelet interactivity. Intravenous inoculation of agg⁺ S. sanguinis strains into rabbits with catheter-induced aortic valve trauma led to larger vegetations, a more severe clinical course, more gross lesions in major organs, and greater mortality than inoculation with an agg⁻ strain or with the agg⁺ strains pretreated with Fab fragments specific for the platelet interactivity phenotype.¹²⁵ Platelet aggregation induced by S. sanguinis in vivo seems to be an important virulence determinant of vegetation development and disease progression. Streptococcal exopolysaccharide production inversely correlates with platelet adhesion while inhibiting aggregation,¹²⁶ indicating that these surface molecules may enhance endocarditis at some pathogenic steps but not at others.

The manner in which S. aureus interacts with platelets in the pathogenesis of IE differs substantially from that of the viridans streptococci. This interaction does not require the presence of specific antistaphylococcal antibody and is not amplified by the platelet Fc receptor.¹²⁷ Platelet–S. aureus interactions for executing aggregation require fibrinogen as a bridging molecule but are independent of the primary platelet fibrinogen-binding site, the glycoprotein IIb/IIIa integrin receptor. In addition, it seems that S. aureus can bind to platelets via platelet-derived von Willebrand factor or directly to von Willebrand factor receptor, at von Willebrand factor–binding domain.^128–130 In addition, platelet surface–expressed gCiqR can serve as a key “docking site” for staphylococci, predominantly through bridging molecules such as protein A and von Willebrand factor.¹³¹ In experimental IE, transposon inactivation of the putative S. aureus platelet-binding adhesin gene resulted in mutants with diminished capacity to adhere to platelets in vitro in either suspension or surface-bound monolayers.¹³² In experimental IE caused by such low-platelet–binding mutants, the induction rates of IE were equivalent to the induction rates of the parental strain, presumably because of the microbe’s ability to attach to damaged endothelium by multiple adhesive mechanisms. However, the capacity of the mutant to persist and proliferate within experimental vegetations and to disseminate hematogenously to the kidneys was markedly impaired in the mutant strain.¹³² This transposon defect was found to reside within the staphylococcal fibrinogen adhesin gene, clumping factor A (clfA).¹³³

Platelets also may play a role in host defense within the cardiac vegetation during IE.¹³⁴ It is underappreciated that platelets can actually phagocytose circulating staphylococci into engulfment vacuoles, in which the organism can persist.¹³⁵ Moreover, after specific exposure to thrombin (which is plentiful at the surface of damaged endothelium), release of α-granule–derived platelet microbicidal proteins (PMPs), or thrombocidins, with bactericidal activity against most gram-positive cocci that cause IE, has been shown.¹³⁶ PMPs appear to be homologues of platelet factor 4,¹³⁷ whereas thrombocidins evolve from the platelet basic peptide lineage and are truncates of the chemokines NAP-2 and CTAP-3.¹³⁸ Although the ability of S. aureus to adhere to and aggregate platelets is a related property, the resistance to PMPs is an independent phenotypic characteristic and a potential virulence factor.¹³⁹ PMPs are low-molecular-weight (8 to 10 kDa) cationic proteins and may act primarily on the bacterial cell membrane or cell wall, synergistically with antibiotics, to kill bacteria. PMPs also have shown fungicidal activity against some yeasts in vitro.¹⁴⁰

Microbial resistance to the cidal activity of PMPs may contribute to the pathogenesis of IE. This hypothesis was supported by a reduction in vegetation weight and bacterial concentration in rabbits with experimental aortic valve S. aureus endocarditis after treatment with aspirin.¹⁴¹ In addition, three studies in experimental IE confirmed the importance of the relationship of PMP resistance and the pathogenesis of IE. In experimental viridans streptococcal IE and S. aureus IE, PMP-resistant strains exhibited an enhanced capacity to persist at sites of valvular damage.^142,143 In addition, S. aureus strains exhibiting the PMP resistance phenotype in vitro were able to proliferate within the vegetation and hematogenously seeded extracardiac foci (kidneys, spleen) to a significantly greater extent than their isogenic counterparts, which were PMP susceptible in vitro.^144,145

Several clinical studies also emphasized the important association between PMP resistance and the pathogenesis of intravascular infections. Bloodstream isolates of viridans streptococci and S. aureus from patients with IE tended to be substantially more resistant to PMPs in vitro.¹⁴³ S. aureus bloodstream isolates arising from an intravascular source (catheter or IE) were significantly more resistant than bloodstream isolates arising from a deep tissue source.¹⁴⁶ Further, PMP-resistant S. aureus bloodstream isolates from patients with IE were significantly more likely to have arisen from an intravascular catheter source.¹⁴⁷ Lastly, recent data have shown fusion of PMP-rich α-granules with platelet engulfment vacuoles noted earlier, implicating the intrinsic PMP-resistance phenotypes of engulfed staphylococci in their ultimate intraplatelet survival outcomes.¹³⁵ Should the engulfed organism be PMP resistant, the platelet could serve as a “Trojan horse” vehicle to disseminate the organism. In contrast, if the phagocytosed staphylococci were PMP susceptible, this mechanism could eliminate the organism from the circulation.¹³⁵

In IE, the bacterial colonies are found beneath the surface of the vegetation (at variable depth, depending on the intracardiac location¹⁴⁸) and infiltration by phagocytic cells is minimal; the vegetation creates an environment of impaired host resistance. These conditions allow for relatively unbridled bacterial growth, resulting in extremely high colony counts of 10⁹ to 10¹¹ bacteria per 1 g of tissue. Bacteria deep within the fibrin matrix have been shown by autoradiography to reach a state of reduced metabolic activity.¹⁴⁹ Studies by Yersin¹⁵⁰ and Meddens¹⁵¹ and their colleagues suggested that impairment of host defenses (e.g., neutropenia, corticosteroids) potentiates progression of the disease when the tricuspid, but not the aortic, valve is involved but is largely dependent on the intracardiac location of the vegetation.¹⁵²

The role of granulocytes within the vegetation is unknown. When vegetation formation is retarded with anticoagulants in experimental animals with IE, the organisms seem to divide on the surface, total bacterial titers are lower, and the clinical disease is more explosive.^153,154 In addition, it has been suggested that phagocytosis of microorganisms by monocytes on or within the vegetation generates tissue thromboplastin formation; thromboplastin then acts as a stimulant to fibrin deposition and growth of the vegetation.¹⁵⁵ However, the best evidence suggests that coagulation activation initiated by tissue factor,¹⁵⁶ with subsequent local thrombus formation, is responsible for the initiation of vegetation growth and persistence on the cardiac valve. It seems that some organisms (e.g., S. aureus) induce tissue factor production by endothelium without the necessity for host cytokines.¹⁵⁷

Many important studies have elucidated the interactions among the invading microbe, the endothelium, and the monocyte in the pathogenesis of IE. After internalization by endothelial cells in vitro, microbes such as S. aureus evoke a potent proinflammatory chemokine response, including, for example, increased expression of interleukin-6 (IL-6) or IL-8 or of monocyte chemotactic peptide.^158,159 This event also is associated with increased expression on the endothelial cell surface of several key adhesion molecules, especially intercellular adhesion molecule 1 (ICAM-1) and vascular cell adhesion molecule 1 (VCAM-1).^106,158,159 Among other cells, monocytes are drawn into this endothelial cell microenvironment; via their appropriate counter-receptors, monocytes can bind avidly to such microbe-activated endothelial cells.¹⁵⁹ Extracellular bacteria circulating in the vascular system then bind directly to the monocyte surface, inducing the release of tissue thromboplastin (tissue factor).^160,161 This latter molecule participates in the catalytic conversion of prothrombin to thrombin, amplifying the procoagulant cascade at the site of endothelial cell colonization and leading to progressive evolution of the vegetative lesion in IE. Several studies have emphasized that organisms with low protease production (e.g., enterococci) are associated with larger, more friable vegetations with an increased tendency for embolization. This property was underscored in a recent elegant animal study by inducing IE with an enterococcus with low proteolytic activity and virulence to analyze the role of host proteases in vegetation growth. Matrix metalloprotease 9, elastase, and plasminogen activators were all present at higher concentrations in septic vegetations. These results suggest that the continuous attractant signals coming from bacterial colonies can result in chronic injury of myocardial tissues by host proteases.

Immunopathologic Factors

IE results in stimulation of humoral and cellular immunity, as manifested by hypergammaglobulinemia, splenomegaly, and the presence of macrophages in the peripheral blood. The possibility that preformed antibody can increase the likelihood of the development of IE was suggested by the spontaneous occurrence of IE in horses receiving repeated immunizations with live pneumococci.¹⁶² It was suggested that these antibodies produced bacterial agglutination in vivo, which increased the chances of valvular colonization. Studies in animals have suggested a protective role for circulating antibody. Rabbits preimmunized with heat-killed streptococci plus Freund’s adjuvant had a significantly higher ID₅₀ than that noted for nonimmunized controls after aortic valve trauma.¹⁶³ Other studies yielded similar results with S. sanguinis, S. mutans, and S. pneumoniae.^164,165 In additional experiments, antibody directed against cell-surface components (including mannan) reduced the adhesion of C. albicans to fibrin and platelets in vitro and reduced IE production in vivo.¹⁶⁶ This effect may depend on the infecting organism, however, because antibody to S. epidermidis or S. aureus did not prevent the development of IE in immunized animals and did not result in reduced bacterial concentrations in infected vegetations or kidneys,¹⁶⁷ perhaps because of the inability of immune sera to enhance opsonophagocytosis of staphylococci. The role of preformed antibody in the pathogenesis of IE is unclear. Intravascular agglutination of bacteria may decrease the frequency of IE by reducing the actual number of circulating organisms, but cross-protection was not achieved by passive transfer of high-titer immune globulin from S. defectivus–immunized rabbits to control animals.¹⁶⁵ Nitrogen mustard–treated immunized rabbits lost their ability to clear S. defectivus efficiently from the circulation, a process partially restored by neutrophil transfusion.¹⁶⁸

The role of the glycocalyx of S. aureus, and of antibodies directed against this exopolysaccharide, in the pathogenesis of IE is controversial. Most experimental studies suggest that microencapsulation of strains by the common capsular types (5 and 8) may mitigate virulence of the organism in IE by obscuring key surface-expressed adhesins involved in colonization or persistence at endovascular damage sites.¹⁶⁹

Several more recent studies suggested a salutary effect of active or passive immunization strategies against this glycocalyx in diminishing the induction, progression, or metastatic infection phases of experimental IE.^170,171 However, large clinical trials have not been able to document salutary outcomes using active or passive immunization strategies directed against the staphylococcal capsule or candidate surface adhesins (e.g., clumping factor).^172–175

Rheumatoid factor (anti-IgG IgM antibody) develops in about 50% of patients with IE of longer than 6 weeks’ duration.¹⁷⁶ Rheumatoid factors were found at the time of admission in 24% of patients with acute staphylococcal IE (<6 weeks’ duration), and the frequency increased to 40% if fever persisted for 2 weeks after the initiation of antibiotic therapy.¹⁷⁷ More than two thirds of the patients became seronegative after 6 weeks of therapy, and two patients with a second episode of acute IE promptly redeveloped positive rheumatoid factors. The titers correlated with the level of hypergammaglobulinemia and decreased with therapy. Rheumatoid factor may play a role in the disease process by blocking IgG opsonic activity (i.e., by reacting with the Fc fragment), stimulating phagocytosis, or accelerating microvascular damage. Rheumatoid factor (IgM) has not been eluted from the immune complex glomerulonephritis associated with IE.¹⁷⁸ Antinuclear antibodies also occur in IE and may contribute to the musculoskeletal manifestations, low-grade fever, or pleuritic pain.¹⁷⁹

Similar to malaria, schistosomiasis, syphilis, kala-azar, and leprosy, IE is associated with a constant intravascular antigenic challenge, and the development of several classes of circulating antibody is not unexpected. Opsonic (IgG), agglutinating (IgG, IgM), and complement-fixing (IgG, IgM) antibodies and cryoglobulins (IgG, IgM, IgA, C3, fibrinogen); various antibodies to bacterial heat-shock proteins; and macroglobulins all have been described in IE.^180–182 Circulating immune complexes have been found in high titers in almost all patients with IE.¹⁸³ Circulating immune complexes are found with increased frequency in connection with a long duration of illness, extravalvular manifestations, hypocomplementemia, and right-sided IE. Levels decrease and become undetectable with successful therapy. Patients with IE and circulating immune complexes may develop a diffuse glomerulonephritis that is analogous to the nephritis seen with infected ventriculoatrial shunts.¹⁸⁴ Corticosteroids have been used in a few patients with glomerulonephritis associated with IE.¹⁸⁵ Immune complexes plus complement are deposited subepithelially along the glomerular basement membrane to form a “lumpy-bumpy” pattern. Immune globulin eluted from these lesions has been shown to cross react with bacterial antigens.¹⁸⁶ In addition, bacterial antigens have been shown within circulating immune complexes.¹⁸⁷

Some of the peripheral manifestations of IE, such as Osler’s nodes, also may result from a deposition of circulating immune complexes. Pathologically, these lesions resemble an acute Arthus reaction. The finding of positive culture aspirates in Osler’s nodes¹⁸⁸ suggests, however, that they may be caused by septic emboli rather than immune complex deposition. In some diffuse purpuric lesions in IE, immune complex deposits (IgG, IgM, and complement) have been shown in the dermal blood vessels by immunofluorescence.¹⁸⁹ Quantitative determinations of serum immune complex concentrations are useful in gauging the response to therapy. Effective treatment leads to a prompt decrease, with eventual disappearance of circulating immune complexes.¹⁹⁰ Conversely, therapeutic failures or relapses are characterized by rising titers or a reappearance of circulating immune complexes.¹⁹¹