- Email: [email protected]
Tricuspid Valve Endocarditis due to Staphylococcus aureus
Tricuspid Valve Endocarditis due to Staphylococcus aureus* Correlation of Two-Dimensional Echocardiography with Clinical Outcome Arnold S. Bayer, M.D., F.C.C.P.;t Ingrid K. Blomquist, M.D.;t Erlaine Bello, M.D.;t Chung-Yin Chiu, M.S.;§ Joel I. Ward, M.D.;II and Leonard E. Ginzton, M.D.~
We studied 53 episodes (51 patients) of tricuspid valvular endocarditis caused by Staphylococcus aureus in a predominantly addicted population and correlated two-dimensional echocardiographic findings with clinical outcome. Thirtyeight episodes with (vs15episodes without) tricuspid vegetations on the two-dimensional echocardiogram were significantly associated with (1) longer duration of fever on therapy (mean of 12.3 days vs 6.8 days, respectively; p
progressive right-sided heart failure (p<0.05 vs patients with normal RVED dimension). Tricuspid vegetations greater than 1.0 cm identified a subset of patients at increased risk for developing clinical right-sided heart failure during the active or convalescent phase of endocarditis (p<0.02 vs patients with tricuspid vegetations less than 1.0 em), An unexpectedly high prevalence of asymptomatic prolapse of the mitral valve was observed in this population (23 of 53 episodes; 43 percent), Detection of tricuspid vegetations in patients with endocarditis due to S aureus is not a primary indication for early surgery, but identifies patients more likely to exhibit short-term and long-term complications of their infection.
Tricuspid valvular endocarditis is essentially a disease of intravenous drug addicts and is caused by Staphylococcus aureus in about 90 percent of the cases.!" Although most studies of addict-associated tricuspid endocarditis report a favorable prognosis with medical therapy alone (cure rates of about 90 to 95 percents"), there remains a subset of such patients who may have a less salutary outcome and are characterized by large tricuspid vegetations (greater than 1 em) on echocardiograms.v' however, in the most recently reported study in this regard," S aureus alone accounted for only about 50 percent of the cases of tricuspid endocarditis. This raised the possibility that the etiologic organism(s) may be as influential as the size of valvular vegetations upon the clinical outcome. The current study was designed to prospectively study the association between two-dimensional echocar-
diographic analyses performed early in the course of tricuspid endocarditis due to S aureus and subsequent clinical complications and outcomes.
*From the Departments of Medicine and Pediatrics, Harbor-UCLA Medical Center, Torrance, CA, and the UCLA School of Medicine, Los Angeles. Supported by the American Heart Association, Greater Los Angeles Affiliate (research grant 853-Gl-l). Presented in part at the national meeting of the American Federation for Clinical Research, San Diego, Calif May 4, 1987. tProfessor of Medicine. :l:Senior Fellow in Infectious Diseases. §Research Associate in Pediatrics. [Associate Professor of Pediatrics. ~Assistant Professor of Medicine. Manuscript received June 4; revision acceJ>ted July 16. Reprint requests: Dr. Bayer, Division oflnfectious Diseases, H arborUCLA Medical Center, 1000 Carson Street, Torrance, CA 90509
MATERIALS AND METHODS
Patients Between January 1980 and January 1987, there were 51 consecutive adult patients with the clinical diagnosis of tricuspid endocarditis due to S aureus who were admitted to the medical ward service of the Harbor-UCLA Medical Center and were prospectively entered into the study. Each patient had at least two sets of positive blood cultures taken from different venous sites, at separate sampling times, that were positive for S aureus. The organism was identified by standard microbiologic techniques." A total of 53 episodes of tricuspid endocarditis due to S aureus occurred in 51 patients. To exclude possible relapses, multiple episodes of endocarditis in the same patient were to be separated in time by at least six months. Fifty-one of the 53 episodes occurred in intravenous drug addicts, one patient with Ebsteins anomaly had two episodes. The five diagnostic criteria for tricuspid endocarditis were (1) compatible symptoms (especially fever, arthralgia, pleurisy, and cough); (2) physical findings suggestive of pneumonitis or pleuritis; (3) radiographic findings compatible with septic pulmonary emboli, occurring in the absence of an extracardiac focus (eg, septic thrombophlebitis), (4) multiple cultures of blood positive for S aureus (more than two sets [four bottlesj), and (5) cardiac auscultatory evidence of tricuspid insufficiency.6,9 All patients entered into the study fulfilled at least four of the five criteria listed (numbers 1 to 4); in addition, in 36 of the 53 episodes (68 percent), a typical murmur of tricuspid insufficiency was noted upon entry into the study. The presence of hematuria, hypocomplementemic renal insufficiency, and positive serum assays for VDRL and rheumatoid factor were often useful in supporting the diagnosis of endocarditis. 10 Also, when CHEST I 93 121 FEBRUARY, 1988
247
available, surgical findings were used to ultimately confirm the diagnosis of endocarditis. Patients with polymicrobial tricuspid endocarditis, as well as those with clinical or echocardiographic evidence of multivalvular endocarditis, were excluded from this analysis. All patients were initially evaluated and followed by members of the Adult Infectious Diseases Consultation Service. Decisions regarding antimicrobial therapy (eg, choices of antibiotics, duration of therapy, dosages, etc), diagnostic work-ups, and need for surgery were made by the attending and house staff physicians of record on the case, without input from the investigative team. Nine episodes included in this current series have been previously published in part in a report of the echocardiographic natural history of tricuspid endocarditis. 11 All entered patients were followed for a minimum of three months after discharge from the hospital.
Echocardwgraphic Studies Patients enrolled in the study had two-dimensional echocardiograms performed within seven days of entry. Many patients had serial echocardiograms performed during their hospitalization; however, since this investigation was focused on the prognostic value of initial two-dimensional echocardiographic findings, only the first echocardiographic results were considered in this analysis. A tricuspid vegetation was defined as a localized discrete mass of reflective echoes attached to this valve." The following two-dimensional echocardiographic parameters were noted: (1) presence or absence of vegetations; (2) size of vegetation in multiple projections, with the largest measured diameter used in the correlation with the clinical findings; (3) right ventricular end-diastolic (RVED) dimension (this measurement was made in the mid-right ventricular end-diastolic minor axis from the apical four-chamber view just beyond the tips of the tricuspid leaflets; 3 cm was determined to be the upper limit of normal for this dimension by one of the investigators [LEG] after analyzing twodimensional echocardiographic data from a large series of normal patients without underlying cardiac disease [unpublished datal): (4) left ventricular end-systolic (LVESV) and end-diastolic volumes (LVEDV) (less than 35 ml and less than 90 ml, respectively, used as the upper limits of normal") and left ventricular ejection fraction (LVEF) (more than 50 percent defined as normalf'), and (5) presence of other associated valvular abnormalities (eg, mitral valve prolapse). The sensitivity of the two-dimensional echocardiogram regarding the size of a tricuspid vegetation is 3 em. In four patients the twodimensional echocardiographic study was technically inadequate for delineating the LVESV: LVEDV: and LVEF: The echocardiographic studies were read and interpreted by one of the authors (LEG) blinded to the patients' names and clinical information. All echocardiograms were performed by the same technical staff. For the first three years of this study, all patients underwent two-dimensional echocardiography with a mechanical sector scanner (Advanced Technology Laboratories, Mark III), while for the last four years of the investigation, all patients were studied with a phased-array scanner (Hewlett-Packard). A high concordance in echocardiographic measurements between these two scanners was validated by comparing ejection fractions with first-pass radionuclide angiography and by comparing calculated left ventricular mass with autopsy-confirmed left ventricular weight in normal dogs and dogs with experimentally induced myocardial infarction. 14
Clinical Complications The patients were carefully evaluated for both short-term and long-term complications. Short-term complications were defined as those that occurred during the patient's hospitalization for antimicrobial therapy for tricuspid endocarditis; these complications generally occurred within the first four weeks after admission. Shortterm complications included prolonged pyrexia, recurrent pulmonary emboli (occurring more than 72 hours after initiation of 248
appropriate antibiotic therapy), and the presence or development of right-Sided heart failure. Prolonged pyrexia was defined as persistent daily fevers higher than 37. SoC (I000F), lasting more than ten days with appropriate antimicrobial therapy," Such fevers were presumed to be related to the endocarditis after careful clinical, radiographic, and microbiologic exclusion of focal metastatic suppuration (particularly empyema and pulmonary abscess), drug fever, and antibiotic-resistant strains of S aureus (eg, "methicillin-resistant" strains"), In addition, the adequacy of serum antimicrobial levels and bactericidal activities were confirmed when indicated. Repetitive pulmonary emboli were ascertained on both clinical grounds (recurrent pleuritis, dyspnea, hypoxemia, or hemoptysis) and radiographic criteria (new peripheral pulmonary infiltrates or characteristic changes on ventilation-perfusion pulmonary scintigraphy, or both). The diagnosis of right-Sided heart failure was made on routine clinical, radiographic and electrocardiographic criteria and was not predicated on two-dimensional echocardiographic findings. Long-term complications were those that occurred after the patient's discharge from the hospital; most of these complications occurred between three and six months after discharge. In this predominantly drug-addicted population, compliance with longterm follow-up efforts was suboptimal. Only about one-third of the patients (17/51) returned to our clinics for follow-up evaluations beyond one year after discharge; these patients were followed for an average of about 14 months after discharge. The long-term complications of importance to this study were (1)development of right-Sided heart failure, (2) need for tricuspid ablation or replacement, and (3) recurrent encodarditis.
Statistical Analyses Associations between the two-dimensional echocardiographic findings and clinical outcomes or complications of episodes of tricuspid endocarditis due to S aureus, as well as the associations of other categorically defined variables, were tested by either the Fisher exact or Xi tests. The correlations between the vegetation's size and febrile morbidity or RVED dimensions were estimated by product-moment correlation coefficients. Multivariate analyses correlating the presence of tricuspid vegetations on the two-dimensional echocardiogram with other parameters were performed by logistic regression models. All statistics were computed by using the BMDP statistical software." RESULTS
Echocardiographic Findings
Thirty-eight (72 percent) of 53 episodes of tricuspid endocarditis were associated with vegetations on twodimensional echocardiograms. Of the two patients with a recurrent episode of tricuspid endocarditis, the two-dimensional echocardiogram demonstrated vegetations on the tricuspid valve during each episode; however, the location and size of the vegetations during each episode of endocarditis were distinctly different. A single vegetation was observed in all but one of the 38 positive echocardiograms. Tricuspid prolapse was noted in five patients, four of whom had tricuspid vegetations on the echocardiogram. No pulmonic valvular vegetations were observed. Of interest, 23 (43 percent) of 53 episodes, all in different patients, had clinically silent mitral valve prolapse also noted on the two-dimensional echocardiogram. None of these latter patients had a mitral valve click or click-murmur noted during their clinical evaluations by house staff and Tricuspid V8Jve Endocarditis due to S auf8U8(Bayer st aI)
attending physicians, nor did these patients demonstrate clinical or echocardiographic evidence of mitral valve endocarditis. Twenty-one (91 percent) of the 23 patients with mitral valve prolapse had vegetations demonstrable on the tricuspid valve, as compared with 17 (59 percent) of 29 patients without mitral valve prolapse (p = 0.05). The maximal size of tricuspid vegetations ranged from 0.4 to 3.7 em, with a mean size of1.26±0.13 em, Twenty-one (55 percent) of the 38 episodes associated with vegetations on the two-dimensional echocardiogram had lesions greater than 1.0 cm. Association of Echocardiographic Findings with Clinical Complications and Outcome Fifty (94 percent) of the 53 episodes of tricuspid endocarditis due to S aureus were cured of active infection by antimicrobial therapy alone (Table 1). Three patients required tricuspid valvular resection during treatment of their active infection, each for prolonged pyrexia (more than three weeks in the absence of other definable causes for the fever). In each instance, valvular resection and histopathologic evaluation confirmed the presence of active valvular and paravalvular infection. Rapid and complete resolution of the fever occurred after surgery. Each of these three patients had vegetations by two-dimensional echocardiograms before surgery. Twenty-two (42 percent) of the 53 episodes were associated with fevers for more than ten days while receiving appropriate therapy without identifiable causes. Of these 22 episodes of prolonged fever, tricuspid vegetations were observed in 20 episodes (91 percent); in contrast, among 31 episodes unassociated with prolonged fever on therapy, vegetations were noted in only 18 (58 percent; p=0.OO9). Episodes of tricuspid endocarditis due to S aureus associated with vegetations on the two-dimensional echocardiogram had a significantly longer duration of fever, as comTable I-Association ofClinical Outcome with Two-Dimensional Echocardiographic Demonstration of Tricuspid Valve Vegetations Echocardiographic Tricuspid Vegetations Clinical Findings
Present
Absent
p Value*
Prolonged fever Recurrent pulmonary emboli Right-sided CHFt During active phase of endocarditis During convalescent phase of endocarditis
20/38 10/38
2/15 2/15
<0.01 NS
5/38
1/15
NS
3/38
IllS
*NS, Not Significantlydifferent. tCHF, Clinically evident congestive heart failure.
NS
pared to those episodes without demonstrable vegetations (12.3 ± 1.6 days vs 6.8 ± 1 days, respectively; p = o. (05). There was no significant correlation between the size of the vegetation (in centimeters) and the duration of fever (days) (r=0.27; p=0.18). The next most frequently observed short-term complication was recurrent pulmonary emboli on therapy, which occurred during 12 (23 percent) of 53 episodes. In ten (83 percent) of these 12 episodes of recurrent pulmonary emboli, vegetations were noted on the twodimensional echocardiogram; however, there was no significant relationship between the presence of vegetations on the echocardiogram and the development of recurrent pulmonary emboli; ten of38 episodes associated with vegetations on the echocardiogram had this syndrome, as compared to two of 15 episodes without demonstrable vegetations (p = 0.26). There was also no significant association between the size of the vegetation and the occurrence of repetitive pulmonary emboli. Six (11 percent) of the 53 episodes of tricuspid endocarditis were associated with the development of clinical evidence of right-sided heart failure during the active phase of infection; five had vegetations on the echocardiogram. Four additional patients among the 17 with long-term clinical follow-up developed evidence of right-sided heart failure during the convalescent phase of their endocarditis; this is exclusive of the patients who underwent tricuspid ablation for recalcitrant endocarditis, since right-sided failure is a common consequence of this procedure." Thus, overall ten (19 percent) of the 53 episodes were associated with the development of right-sided heart failure during the active or convalescent phase of infection. Eight of these ten episodes had vegetations shown on the twodimensional echocardiogram. Increased RVED dimension (more than 3.0 em) was noted on the initial two-dimensional echocardiogram during 25 (47 percent) of the 53 episodes of tricuspid endocarditis, although only six (24 percent) of these 25 patients had clinical evidence of right-sided failure. There was a significant association between the presence of vegetations on the echocardiogram and increased RVED dimensions (Table 2). Twenty-three (92 Table 2-Association ofIncreaee« in RVED Dimensions on Two-Dimenaional Echocardiogram with Presence of Tricuspid Vegetations and Need for Eventual Tricuspid Surgery Incidence Parameter
Increased RVED*
Normal RVED
p Value
Tricuspid vegetation Need for tricuspid surgery
23/25 5/25
2/28 0/28
<0.0005 0.05
*Increased right ventricular end-diastolic dimension (>3.0 em), CHEST / 93 / 2 / FEBRUARY. 1988
248
percent) of the 25 episodes with increases of RVED dimension on the echocardiogram also had tricuspid vegetations, as compared to only two (7 percent) of 28 episodes with normal RVED dimension (p<0.OOO5). There was no direct correlation between maximal size of the vegetation and RVED dimensions (r = 0.1; p = 0.63). Of the five patients who required tricuspid resection (three for prolonged fever despite appropriate antibiotic therapy and two for progressive rightsided heart failure during convalescence), all had an increased RVED dimension on the initial two-dimensional echocardiogram. Overall, five (20 percent) of25 patient-episodes associated with increases in RVED diameter on the initial echocardiogram resulted in eventual tricuspid resection, as compared to none of the 28 episodes without demonstrable increases in RVED dimension (p<0.05). Upon multivariate analysis, the presence of tricuspid vegetations on the two-dimensional echocardiogram was most significantly associated with development of prolonged fever despite appropriate antimicrobial therapy (p<0.05). None of the following parameters was independently associated with the presence of tricuspid vegetations: age or sex of patient; presence of right-sided congestive heart failure or peripheral stigmata of endocarditis on entry into the study; or development of repetitive pulmonary emboli or new-onset congestive heart failure during antimicrobial therapy.
Clinical-Echocardiographic Associations in Patients with Large Vegetations The risks of prolonged fever and of repetitive pulmonary emboli and the development of right-sided heart failure were not significantly different in episodes with or without vegetations larger than 1.0 em. Moreover, episodes with vegetations larger than 1.0 em had only a slightly longer duration of fever, as compared to episodes with smaller vegetations (13days vs 11.3 days, respectively; p>0.10); however, in the subgroup of episodes with vegetations larger than 1.0 em, significantly more episodes of tricuspid endocarditis were associated with development of right-sided heart failure during either the active or convalescent phase of infection than episodes in which smaller vegetations were noted (p = 0.02). No other grouping of clinical complications was significantly associated with vegetations larger than 1.0 cm.
Mitral Valve Prolapse To determine that the high prevalence of mitral valve prolapse was a "true-positive" and related to structural abnormalities of the mitral valve apparatus, rather than associated with disproportionately small ven tricles (Uvalvulo-ven tricular dispropor tion?"), values for LVESV and LVEDV were compared in 250
Table 3-Left Ventricular Function by Two-DimtmBional Echocardiography in Patients with Tricuspid Endocarditia with and without Mitral Valve Prolapae Mitral Valvular Prolapse Parameter
Present
Absent*
p Valuet
Increased LVEDV (>90 ml) Increased LVESV (>35 ml) Decreased LVEF «50 percent)
17/2:3 is/2:3 4/23
12126
0.05 <0.02 NS
8/26 5/26
*Four additional patients without mitral valvular prolapse had suboptimal studies for LVED~ LVES~ and LVEF and are not included in this analysis. tNS, Not significant.
patients with and without prolapse. Of interest, there
was a significant association between increases in LVESV and LVEDV and the presence of mitral valve
prolapse on the two-dimensional echocardiogram (Table 3); for example, 15 (65 percent) of23 episodes with mitral valve prolapse had increases in LVES~ as compared to only eight (31percent) of 26 episodes without mitral valve prolapse (p<0.02). Similarly, 17 (74 percent) of 23 episodes with, compared with 12 (46 percent) of 26 episodes without, mitral valve prolapse were associated with increased LVEDV (p=0.05). Moreover, values for LVEF in patients with and without mitral valve prolapse were similar. These data suggested that the mitral valve prolapse seen in patients with tricuspid endocarditis was caused by structural abnormalities of the mitral valve apparatus (rather than disproportionately small left ventricular size) and was associated with clinically silent mitral regurgitation. DISCUSSION
Over the past 15 years, the echocardiogram has developed into a valuable noninvasive tool for the diagnosis and management of patients with infective endocarditis. 20 The role of this technique in evaluating patients with infective endocarditis has been expanded even further since inception of the two-dimensional scanning techniques. Whereas the earlier M-mode scanners were limited to visualization of the mitral valve and only two aortic, one pulmonic, and one tricuspid leaflet, the two-dimensional scanning technique provided more complete delineation of all valvular leaflets. 21 Moreover; the size, shape, mobility, and points of attachment of vegetations are also better seen on two-dimensional echocardiographic studies than with the M-mode technique. In addition, vegetations on previously damaged valves are more easily identified by the two-dimensional scanner. A number of investigators have associated both the echocardiographic detection and size of vegetations on aortic and mitral valves with an increased risk of complicaTrIcuspid Valve Endocarditis due to S 8ureus (Bayerat 8/)
tions, particularly major peripheral emboli, congestive heart failure, frank valve disruption, and need for valvular replacement surgery. 7,22-25 In contrast, in patients with right-sided infective endocarditis, the prognostic implications of the size of the vegetation and detection on the two-dimensional echocardiogram have not been well delineated. Robbins et al" have recently reported a series of21 patients with right-sided valvular infective endocarditis and demonstrated that patients with large vegetations on the two-dimensional echocardiogram had a significantly worse prognosis with medical therapy. Thus, four (36 percent) of their 11 patients with vegetations greater than 1.0 em required surgery for prolonged pyrexia while receiving appropriate antibiotic therapy, as compared to none of eight patients with smaller vegetations. Some interpretive problems exist with this study. It was not clear whether the clinical evaluations were performed prospectively or retrospectively by chart review, and whether these were done by the same investigative team at the two institutions. Importantly, only 11 (48 percent) of the 23 episodes were caused by S aureus alone, with the remaining episodes being polymicrobial, caused by other aerobic Gram-positive cocci, or "culture-negative." Lastly, four (17 percent) of the 23 episodes, including one of the cases requiring tricuspid valve surgery, involved either the pulmonic valve alone (three cases) or the mitral valve in addition to the tricuspid valve. The multi-institutional aspects of this study, as well as the bacteriologic and valve-site diversities in their patients make interpretation of their observations problematic. Wong et al? also attempted to correlate larger tricuspid vegetations on M-mode or two-dimensional echocardiograms with a less favorable clinical outcome; however, only seven of their ten cases of tricuspid endocarditis were caused by S aureus; of note, two of the three patients who required tricuspid surgery had endocarditis caused by organisms other than S aureus (5 epidermidis and Pseudomonas aeruginosa in one case each). We designed the current study to include only patients with tricuspid endocarditis due to S aureus in order to avoid the interpretive difficulties associated with bacteriologic and valve-site diversities. Several interesting observations resulted from our investigation. Most patients (50/53; 94 percent) in our study were bacteriologically cured of active tricuspid endocarditis with antimicrobial therapy alone. Patients with tricuspid vegetations noted on two-dimensional echocardiograms remained febrile significantly longer despite appropriate therapy (about two weeks, mean) as compared to patients without demonstrable vegetations (about one week, mean); however, neither the actual size of the vegetation nor the presence of
vegetations larger than 1.0 cm was significantly correlated with the duration offever. Also, in contrast to leftsided infection, where the echocardiographic demonstration ofvegetations is associated with increased risks of major embolic episodes," this study of tricuspid endocarditis could not document such an association vis-a-vis recurrent pulmonary emboli while on appropriate therapy. Clinically evident right-sided heart failure was a relatively uncommon consequence of tricuspid endocarditis due to S aureus in our population, occurring in less than 20 percent of the episodes in either the active or convalescent phases of infection; however, the presence of tricuspid vegetations was significantly correlated with increases in RVED dimensions on initial two-dimensional echocardiograms. This finding suggests that patients with tricuspid endocarditis due to S aureus and vegetations on the two-dimensional echocardiogram are at Significantly increased risk for incipient right-sided heart failure. This presumption is further supported in our study by the fact that Significantly more patients with increased RVED dimensions (20 percent) eventually required tricuspid surgery than their counterparts with normal RVED dimensions (zero). The presence of vegetations larger than 1.0 em on two-dimensional echocardiograms in our patients with tricuspid endocarditis was not Significantly associated with any individual prognostic parameter on univariate analyses; however, the presence of such larger vegetations did significantly correlate with an increased propensity for such patients to develop clinical evidence of right-sided heart failure during either the active or convalescent phase of their infection. One must interpret this correlation with some caution, as the long-term followup in our study was not optimal. Only about onethird of the patients were followed for more than one year after discharge from the hospital. It is thus entirely possible that the frequency of developing clinically evident right-sided heart failure following tricuspid endocarditis due to 5 aureus is higher than we documented (about 20 percent). The high incidence of echocardiographically defined mitral valve prolapse (23/53; 43 percent) in this study was unexpected. No episodes were associated with clinical evidence of either the "click-murmur" syndrome" or mitral valve endocarditis on initial or followup evaluation. It is known that the prevalence of mitral valve prolapse on the two-dimensional echocardiogram is not different in uninfected drug addicts as compared with normal patients. Siegel et al26 from our own institution demonstrated that 6 percent of addicts actively using intravenous narcotic drugs, but without evidence of endocarditis, had mitral valve prolapse on two-dimensional echocardiograms. The echocardiographic demonstration of mitral valve prolapse appeared to represent structural alterations of the mitral CHEST I 93 I 2 I FEBRUARY, 1988
251
valve apparatus in our patients, rather than possible "false-positive" valve prolapse related to small ventricular dimensions. This was supported by finding significant increases in LVESV and LVEDV in patients with (vs without) mitral valve prolapse in our study, as well as also observing essentially normal LVEF in this population. Why our predominantly addicted population with tricuspid endocarditis due to S aureus has such a high prevalence of underlying mitral valve prolapse by two-dimensional echocardiograms remains unclear. One possibility is that patients with mitral valve prolapse also have a more generalized myxomatous degeneration of other valves, including the tricuspid valve. This has, in fact, been demonstrated in postmortem studies of patients with mitral valve prolapse. 27 It is also well documented that addicts are exposed frequently to S aureus bacteremia from intravenous narcotic self-injections, performed under suboptimal hygienic conditions, through cutaneous sites which are heavily colonized with staphylococci.28-30 Thus, it seems reasonable that the high prevalence of mitral valve prolapse in our study represents a "selection" of staphylococcemic patients with more global myxomatous degeneration of the atrioventricular valves predisposing to tricuspid endocarditis; however, this theory remains speculative. In summary, the demonstration of tricuspid vegetations on two-dimensional echocardiograms in patients with right-sided endocarditis due to S aureus identifies a subset of patients that is (1) likely to exhibit a longer time to defervescence with appropriate medical therapy, (2) likely to have findings of subclinical right-sided heart failure, and (3) at increased risk for eventual tricuspid surgery. Patients with vegetations larger than 1.0 em appear at increased risk for development of clinically evident right-sided heart failure during the active or convalescent phases of their infection. Our analysis confirms that the vast majority of patients with tricuspid endocarditis due to S aureus are cured medically without serious short-term or long-term sequelae, regardless of the presence of tricuspid vegetations on the two-dimensional echocardiogram, and that this latter finding alone should not dictate early empiric valvular surgery; however, the demonstration of tricuspid vegetations in this population does identify a subset of patients who require careful clinical and echocardiographic follow-up for possible long-term complications of endocarditis due to S aureus. A similar position has been recently stressed by Dinubile." ACKNOWLEDGMENTS: We thank Susan Partridge, R.N., for excellent technical assistance throughout this study.
REFERENCES 1 Abrams B, Sklaver A, Hoffman l: Greenman R. Single or combination therapy of staphylococcal endocarditis in intravenous drug abusers. Ann Intern Med 1979; 90:789-791 2 Menda KB, Gorbach SL. Favorable experience with bacterial
252
endocarditis in drug addicts. Am Heart J 1973; 78:25-32 3 Chambers HF, Korzeniowski OM, Sande MA. Staphylococcus aureus endocarditis: clinical manifestations in addicts and nonaddicts. Medicine 1983; 62:170-77 4 EI-Khatid MR, Wilson FM, Lerner AM. Characteristics of bacterial endocarditis in heroin addicts in Detroit. Am J Med Sci 1976; 271:197-201 5 Banks l: Fletcher R, Ali N. Infective endocarditis in heroin addicts. Am J Med 1973; 55:444-51 6 Robbins MJ, Frater RWM, Soeiro R, Frishman WH, Strom JA. Influence of vegetation size on clinical outcome of right-sided infective endocarditis. Am J Med 1986; 80:165-71 7 Wong 0, Chandraratna PAN, Wishnow RM, Dusitnanond ~ Nimalasuriya A. Clinical implications of large vegetations in infectious endocarditis. Arch Intern Med 1983; 143:1874-77 8 Baird-Parker AC. Gram-positive cocci. In: Buchanan RE, Gibbons NE, eds. Bergey's Manual of determinative bacteriology, 8th ed. Baltimore: Williams and Wilkins, 1974:478-528 9 Bayer AS, Lam K, Ginzton L, Norman DC, Chiu CY, Ward JI. Staphylococcus aureus bacteremia: clinical, serologic and echocardiographic findings in patients with and without endocarditis. Arch Intern Med 1987; 147:457-62 10 Bayer AS. Infective endocarditis with renal involvement. In: Glassock RJ, ed. Current therapy in nephrology and hypertension. Philadelphia: BC Decker, Inc, 1985:135-38 11 Ginzton LE, Siegel RJ, Criley JM. Natural history of tricuspid valve endocarditis: a two-dimensional echocardiographic study. Am J Cardiol 1982; 49:1853-59 12 Stratton JR, Werner JA, Pearlman AJ, Junko CJ, Kliman S, Jackson MC. Bacteremia and the heart: serial echocardiographic findings in 80 patients with documented or suspected bacteremia. Am J Med 1982; 73:851-58 13 Feigenbaum H. Echocardiography. Philadelphia: Lea and Febiger, 1986:636 14 Ginzton LE, Brazelton M, Delaney ~ Adomian G, Garner 0, Laks MM. Quantitation of two-dimensional echocardiography: day-to-day variability of left ventricular mass in the conscious dog. Clin Res 1983; 31:9A 15 Levinson ME. Infective endocarditis: response to therapy. In: Kaye D, ed. Infective endocarditis. Baltimore: University Park Press, 1976:185-89 16 Saravolatz LD, Pohlod OJ, Arking LM. Community-acquired methicillin-resistant Staphylococcusaureus infections. Ann Intern Med 1982; 97:325-29 17 Dixon WJ. BMDP statistical software. Berkeley, Calif: University of California Press, 1983 18 Arbulu A, Asfaw I. Tricuspid valvulectomy without prosthetic replacement: ten years of clinical experience. J Thorac Cardiovasc Surg 1981; 82:684-91 19 Ginzton LE. Prolapsed mitral leaflet syndrome. In: Shine KI, ed. New York: John Wiley and Sons, 1983:1-20 20 Popp RL. Echocardiography and infectious endocarditis. In: Remington JS, Swartz MN, eds. Current clinical topics in infectious diseases. New York: McGraw-Hill Book Co, 1983:98110 21 Mintz GS, Kotler MN. Clinical value and limitations of echocardiography: its use in the study of patients with infectious endocarditis. Arch Intern Med 1980; 140:1022-26 22 Stewart JA, Silimperi D, Harris ~ Wise NK, Fraker TD, Kisslo JA. Echocardiographic documentation of vegetative lesions in infectious endocarditis: clinical implications. Circulation 1980; 61:374-80 23 Mintz GS, Kotler MN, Segal BL, Parry W A comparison of twodimensional and M-mode echocardiography in the evaluation of patients with infective endocarditis. Am J Cardiol 1979; 43: 738-44 24 Wann LS, Hallam CC, Dillon jc, Weyman AE, Feigenbaum H. Tricuspid Valve Endocarditis due to S aureus(Bayer et aJ)
25
26 27 28
Comparison of M-mode and cross-sectional echocardiography in infective endocarditis. Circulation 1979; 60:728-733 Davis RS, Strom jA, Frishman WH, Becker R, Matsumoto M, LeJemtel TH, et ale The demonstration of vegetations by echocardiography in bacterial endocarditis: an indication for early surgical intervention. Am J Med 1980; 69:57-63 Siegel RJ, Guze RA, Beland K, Guze L, Criley JM. Twodimensional echoeardiographic findings in drug addicts. J Cardiovasc Ultrasonography 1982; 1:133-34 Pomerance A. Ballooning deformity (mucoid degeneration) of atrio-ventricular valves. Br Heart J 1969; 31:343-51 Crane LR, Levine D~ Zervos MJ, Cummings G. Bacteremia in
narcotic addicts at the Detroit Medical Center: microbiology, epidemiology, risk factors and empiric therapy. Rev Infect Dis
1986; 8:364-73
29 TuazonCU, Sheagren IN. Increased rate of carriage of Staphylococcus aurew among narcotic addicts. J Infect Dis 1974; 129: 725-27 30 TuazonCU, Sheagren IN. Staphylococcal endocarditis in parenteral drug abusers: source of the organism. Ann Intern Med 1975; 129:327-29 31 DiNubile M. Surgery for addiction-related tricuspid valve endocarditis: caveat emptor. Am J Moo 1987; 82:811-13
Current Concepts in Asthma, Allergy and Pulmonary Disease Beth Israel Medical Center and the Page and William Black Post-Graduate School of Medicine of the Mount Sinai School of Medicine, will present this program at Beth Israel Medical Center; April 29. For information, contact Beverley R. Baptiste, Medical Education, Beth Israel Medical Center; First Avenue at 16th Street, New York 10003 (212:420-2849).
Critical Care Medicine '88 The Orlando Regional Medical Center will present this program April 6-9 at the Buena Vista Palace Hotel, Lake Buena Vista, Florida. For information, contact Patti L. Devlin, CME Coordinator, Orlando Regional Medical Center, 1414 South Kuhl Avenue, Orlando 32806-2093 (305:841-5144).
CHEST I 93 121 FEBRUARY. 1988
253
We studied 53 episodes (51 patients) of tricuspid valvular endocarditis caused by Staphylococcus aureus in a predominantly addicted population and correlated two-dimensional echocardiographic findings with clinical outcome. Thirtyeight episodes with (vs15episodes without) tricuspid vegetations on the two-dimensional echocardiogram were significantly associated with (1) longer duration of fever on therapy (mean of 12.3 days vs 6.8 days, respectively; p
progressive right-sided heart failure (p<0.05 vs patients with normal RVED dimension). Tricuspid vegetations greater than 1.0 cm identified a subset of patients at increased risk for developing clinical right-sided heart failure during the active or convalescent phase of endocarditis (p<0.02 vs patients with tricuspid vegetations less than 1.0 em), An unexpectedly high prevalence of asymptomatic prolapse of the mitral valve was observed in this population (23 of 53 episodes; 43 percent), Detection of tricuspid vegetations in patients with endocarditis due to S aureus is not a primary indication for early surgery, but identifies patients more likely to exhibit short-term and long-term complications of their infection.
Tricuspid valvular endocarditis is essentially a disease of intravenous drug addicts and is caused by Staphylococcus aureus in about 90 percent of the cases.!" Although most studies of addict-associated tricuspid endocarditis report a favorable prognosis with medical therapy alone (cure rates of about 90 to 95 percents"), there remains a subset of such patients who may have a less salutary outcome and are characterized by large tricuspid vegetations (greater than 1 em) on echocardiograms.v' however, in the most recently reported study in this regard," S aureus alone accounted for only about 50 percent of the cases of tricuspid endocarditis. This raised the possibility that the etiologic organism(s) may be as influential as the size of valvular vegetations upon the clinical outcome. The current study was designed to prospectively study the association between two-dimensional echocar-
diographic analyses performed early in the course of tricuspid endocarditis due to S aureus and subsequent clinical complications and outcomes.
*From the Departments of Medicine and Pediatrics, Harbor-UCLA Medical Center, Torrance, CA, and the UCLA School of Medicine, Los Angeles. Supported by the American Heart Association, Greater Los Angeles Affiliate (research grant 853-Gl-l). Presented in part at the national meeting of the American Federation for Clinical Research, San Diego, Calif May 4, 1987. tProfessor of Medicine. :l:Senior Fellow in Infectious Diseases. §Research Associate in Pediatrics. [Associate Professor of Pediatrics. ~Assistant Professor of Medicine. Manuscript received June 4; revision acceJ>ted July 16. Reprint requests: Dr. Bayer, Division oflnfectious Diseases, H arborUCLA Medical Center, 1000 Carson Street, Torrance, CA 90509
MATERIALS AND METHODS
Patients Between January 1980 and January 1987, there were 51 consecutive adult patients with the clinical diagnosis of tricuspid endocarditis due to S aureus who were admitted to the medical ward service of the Harbor-UCLA Medical Center and were prospectively entered into the study. Each patient had at least two sets of positive blood cultures taken from different venous sites, at separate sampling times, that were positive for S aureus. The organism was identified by standard microbiologic techniques." A total of 53 episodes of tricuspid endocarditis due to S aureus occurred in 51 patients. To exclude possible relapses, multiple episodes of endocarditis in the same patient were to be separated in time by at least six months. Fifty-one of the 53 episodes occurred in intravenous drug addicts, one patient with Ebsteins anomaly had two episodes. The five diagnostic criteria for tricuspid endocarditis were (1) compatible symptoms (especially fever, arthralgia, pleurisy, and cough); (2) physical findings suggestive of pneumonitis or pleuritis; (3) radiographic findings compatible with septic pulmonary emboli, occurring in the absence of an extracardiac focus (eg, septic thrombophlebitis), (4) multiple cultures of blood positive for S aureus (more than two sets [four bottlesj), and (5) cardiac auscultatory evidence of tricuspid insufficiency.6,9 All patients entered into the study fulfilled at least four of the five criteria listed (numbers 1 to 4); in addition, in 36 of the 53 episodes (68 percent), a typical murmur of tricuspid insufficiency was noted upon entry into the study. The presence of hematuria, hypocomplementemic renal insufficiency, and positive serum assays for VDRL and rheumatoid factor were often useful in supporting the diagnosis of endocarditis. 10 Also, when CHEST I 93 121 FEBRUARY, 1988
247
available, surgical findings were used to ultimately confirm the diagnosis of endocarditis. Patients with polymicrobial tricuspid endocarditis, as well as those with clinical or echocardiographic evidence of multivalvular endocarditis, were excluded from this analysis. All patients were initially evaluated and followed by members of the Adult Infectious Diseases Consultation Service. Decisions regarding antimicrobial therapy (eg, choices of antibiotics, duration of therapy, dosages, etc), diagnostic work-ups, and need for surgery were made by the attending and house staff physicians of record on the case, without input from the investigative team. Nine episodes included in this current series have been previously published in part in a report of the echocardiographic natural history of tricuspid endocarditis. 11 All entered patients were followed for a minimum of three months after discharge from the hospital.
Echocardwgraphic Studies Patients enrolled in the study had two-dimensional echocardiograms performed within seven days of entry. Many patients had serial echocardiograms performed during their hospitalization; however, since this investigation was focused on the prognostic value of initial two-dimensional echocardiographic findings, only the first echocardiographic results were considered in this analysis. A tricuspid vegetation was defined as a localized discrete mass of reflective echoes attached to this valve." The following two-dimensional echocardiographic parameters were noted: (1) presence or absence of vegetations; (2) size of vegetation in multiple projections, with the largest measured diameter used in the correlation with the clinical findings; (3) right ventricular end-diastolic (RVED) dimension (this measurement was made in the mid-right ventricular end-diastolic minor axis from the apical four-chamber view just beyond the tips of the tricuspid leaflets; 3 cm was determined to be the upper limit of normal for this dimension by one of the investigators [LEG] after analyzing twodimensional echocardiographic data from a large series of normal patients without underlying cardiac disease [unpublished datal): (4) left ventricular end-systolic (LVESV) and end-diastolic volumes (LVEDV) (less than 35 ml and less than 90 ml, respectively, used as the upper limits of normal") and left ventricular ejection fraction (LVEF) (more than 50 percent defined as normalf'), and (5) presence of other associated valvular abnormalities (eg, mitral valve prolapse). The sensitivity of the two-dimensional echocardiogram regarding the size of a tricuspid vegetation is 3 em. In four patients the twodimensional echocardiographic study was technically inadequate for delineating the LVESV: LVEDV: and LVEF: The echocardiographic studies were read and interpreted by one of the authors (LEG) blinded to the patients' names and clinical information. All echocardiograms were performed by the same technical staff. For the first three years of this study, all patients underwent two-dimensional echocardiography with a mechanical sector scanner (Advanced Technology Laboratories, Mark III), while for the last four years of the investigation, all patients were studied with a phased-array scanner (Hewlett-Packard). A high concordance in echocardiographic measurements between these two scanners was validated by comparing ejection fractions with first-pass radionuclide angiography and by comparing calculated left ventricular mass with autopsy-confirmed left ventricular weight in normal dogs and dogs with experimentally induced myocardial infarction. 14
Clinical Complications The patients were carefully evaluated for both short-term and long-term complications. Short-term complications were defined as those that occurred during the patient's hospitalization for antimicrobial therapy for tricuspid endocarditis; these complications generally occurred within the first four weeks after admission. Shortterm complications included prolonged pyrexia, recurrent pulmonary emboli (occurring more than 72 hours after initiation of 248
appropriate antibiotic therapy), and the presence or development of right-Sided heart failure. Prolonged pyrexia was defined as persistent daily fevers higher than 37. SoC (I000F), lasting more than ten days with appropriate antimicrobial therapy," Such fevers were presumed to be related to the endocarditis after careful clinical, radiographic, and microbiologic exclusion of focal metastatic suppuration (particularly empyema and pulmonary abscess), drug fever, and antibiotic-resistant strains of S aureus (eg, "methicillin-resistant" strains"), In addition, the adequacy of serum antimicrobial levels and bactericidal activities were confirmed when indicated. Repetitive pulmonary emboli were ascertained on both clinical grounds (recurrent pleuritis, dyspnea, hypoxemia, or hemoptysis) and radiographic criteria (new peripheral pulmonary infiltrates or characteristic changes on ventilation-perfusion pulmonary scintigraphy, or both). The diagnosis of right-Sided heart failure was made on routine clinical, radiographic and electrocardiographic criteria and was not predicated on two-dimensional echocardiographic findings. Long-term complications were those that occurred after the patient's discharge from the hospital; most of these complications occurred between three and six months after discharge. In this predominantly drug-addicted population, compliance with longterm follow-up efforts was suboptimal. Only about one-third of the patients (17/51) returned to our clinics for follow-up evaluations beyond one year after discharge; these patients were followed for an average of about 14 months after discharge. The long-term complications of importance to this study were (1)development of right-Sided heart failure, (2) need for tricuspid ablation or replacement, and (3) recurrent encodarditis.
Statistical Analyses Associations between the two-dimensional echocardiographic findings and clinical outcomes or complications of episodes of tricuspid endocarditis due to S aureus, as well as the associations of other categorically defined variables, were tested by either the Fisher exact or Xi tests. The correlations between the vegetation's size and febrile morbidity or RVED dimensions were estimated by product-moment correlation coefficients. Multivariate analyses correlating the presence of tricuspid vegetations on the two-dimensional echocardiogram with other parameters were performed by logistic regression models. All statistics were computed by using the BMDP statistical software." RESULTS
Echocardiographic Findings
Thirty-eight (72 percent) of 53 episodes of tricuspid endocarditis were associated with vegetations on twodimensional echocardiograms. Of the two patients with a recurrent episode of tricuspid endocarditis, the two-dimensional echocardiogram demonstrated vegetations on the tricuspid valve during each episode; however, the location and size of the vegetations during each episode of endocarditis were distinctly different. A single vegetation was observed in all but one of the 38 positive echocardiograms. Tricuspid prolapse was noted in five patients, four of whom had tricuspid vegetations on the echocardiogram. No pulmonic valvular vegetations were observed. Of interest, 23 (43 percent) of 53 episodes, all in different patients, had clinically silent mitral valve prolapse also noted on the two-dimensional echocardiogram. None of these latter patients had a mitral valve click or click-murmur noted during their clinical evaluations by house staff and Tricuspid V8Jve Endocarditis due to S auf8U8(Bayer st aI)
attending physicians, nor did these patients demonstrate clinical or echocardiographic evidence of mitral valve endocarditis. Twenty-one (91 percent) of the 23 patients with mitral valve prolapse had vegetations demonstrable on the tricuspid valve, as compared with 17 (59 percent) of 29 patients without mitral valve prolapse (p = 0.05). The maximal size of tricuspid vegetations ranged from 0.4 to 3.7 em, with a mean size of1.26±0.13 em, Twenty-one (55 percent) of the 38 episodes associated with vegetations on the two-dimensional echocardiogram had lesions greater than 1.0 cm. Association of Echocardiographic Findings with Clinical Complications and Outcome Fifty (94 percent) of the 53 episodes of tricuspid endocarditis due to S aureus were cured of active infection by antimicrobial therapy alone (Table 1). Three patients required tricuspid valvular resection during treatment of their active infection, each for prolonged pyrexia (more than three weeks in the absence of other definable causes for the fever). In each instance, valvular resection and histopathologic evaluation confirmed the presence of active valvular and paravalvular infection. Rapid and complete resolution of the fever occurred after surgery. Each of these three patients had vegetations by two-dimensional echocardiograms before surgery. Twenty-two (42 percent) of the 53 episodes were associated with fevers for more than ten days while receiving appropriate therapy without identifiable causes. Of these 22 episodes of prolonged fever, tricuspid vegetations were observed in 20 episodes (91 percent); in contrast, among 31 episodes unassociated with prolonged fever on therapy, vegetations were noted in only 18 (58 percent; p=0.OO9). Episodes of tricuspid endocarditis due to S aureus associated with vegetations on the two-dimensional echocardiogram had a significantly longer duration of fever, as comTable I-Association ofClinical Outcome with Two-Dimensional Echocardiographic Demonstration of Tricuspid Valve Vegetations Echocardiographic Tricuspid Vegetations Clinical Findings
Present
Absent
p Value*
Prolonged fever Recurrent pulmonary emboli Right-sided CHFt During active phase of endocarditis During convalescent phase of endocarditis
20/38 10/38
2/15 2/15
<0.01 NS
5/38
1/15
NS
3/38
IllS
*NS, Not Significantlydifferent. tCHF, Clinically evident congestive heart failure.
NS
pared to those episodes without demonstrable vegetations (12.3 ± 1.6 days vs 6.8 ± 1 days, respectively; p = o. (05). There was no significant correlation between the size of the vegetation (in centimeters) and the duration of fever (days) (r=0.27; p=0.18). The next most frequently observed short-term complication was recurrent pulmonary emboli on therapy, which occurred during 12 (23 percent) of 53 episodes. In ten (83 percent) of these 12 episodes of recurrent pulmonary emboli, vegetations were noted on the twodimensional echocardiogram; however, there was no significant relationship between the presence of vegetations on the echocardiogram and the development of recurrent pulmonary emboli; ten of38 episodes associated with vegetations on the echocardiogram had this syndrome, as compared to two of 15 episodes without demonstrable vegetations (p = 0.26). There was also no significant association between the size of the vegetation and the occurrence of repetitive pulmonary emboli. Six (11 percent) of the 53 episodes of tricuspid endocarditis were associated with the development of clinical evidence of right-sided heart failure during the active phase of infection; five had vegetations on the echocardiogram. Four additional patients among the 17 with long-term clinical follow-up developed evidence of right-sided heart failure during the convalescent phase of their endocarditis; this is exclusive of the patients who underwent tricuspid ablation for recalcitrant endocarditis, since right-sided failure is a common consequence of this procedure." Thus, overall ten (19 percent) of the 53 episodes were associated with the development of right-sided heart failure during the active or convalescent phase of infection. Eight of these ten episodes had vegetations shown on the twodimensional echocardiogram. Increased RVED dimension (more than 3.0 em) was noted on the initial two-dimensional echocardiogram during 25 (47 percent) of the 53 episodes of tricuspid endocarditis, although only six (24 percent) of these 25 patients had clinical evidence of right-sided failure. There was a significant association between the presence of vegetations on the echocardiogram and increased RVED dimensions (Table 2). Twenty-three (92 Table 2-Association ofIncreaee« in RVED Dimensions on Two-Dimenaional Echocardiogram with Presence of Tricuspid Vegetations and Need for Eventual Tricuspid Surgery Incidence Parameter
Increased RVED*
Normal RVED
p Value
Tricuspid vegetation Need for tricuspid surgery
23/25 5/25
2/28 0/28
<0.0005 0.05
*Increased right ventricular end-diastolic dimension (>3.0 em), CHEST / 93 / 2 / FEBRUARY. 1988
248
percent) of the 25 episodes with increases of RVED dimension on the echocardiogram also had tricuspid vegetations, as compared to only two (7 percent) of 28 episodes with normal RVED dimension (p<0.OOO5). There was no direct correlation between maximal size of the vegetation and RVED dimensions (r = 0.1; p = 0.63). Of the five patients who required tricuspid resection (three for prolonged fever despite appropriate antibiotic therapy and two for progressive rightsided heart failure during convalescence), all had an increased RVED dimension on the initial two-dimensional echocardiogram. Overall, five (20 percent) of25 patient-episodes associated with increases in RVED diameter on the initial echocardiogram resulted in eventual tricuspid resection, as compared to none of the 28 episodes without demonstrable increases in RVED dimension (p<0.05). Upon multivariate analysis, the presence of tricuspid vegetations on the two-dimensional echocardiogram was most significantly associated with development of prolonged fever despite appropriate antimicrobial therapy (p<0.05). None of the following parameters was independently associated with the presence of tricuspid vegetations: age or sex of patient; presence of right-sided congestive heart failure or peripheral stigmata of endocarditis on entry into the study; or development of repetitive pulmonary emboli or new-onset congestive heart failure during antimicrobial therapy.
Clinical-Echocardiographic Associations in Patients with Large Vegetations The risks of prolonged fever and of repetitive pulmonary emboli and the development of right-sided heart failure were not significantly different in episodes with or without vegetations larger than 1.0 em. Moreover, episodes with vegetations larger than 1.0 em had only a slightly longer duration of fever, as compared to episodes with smaller vegetations (13days vs 11.3 days, respectively; p>0.10); however, in the subgroup of episodes with vegetations larger than 1.0 em, significantly more episodes of tricuspid endocarditis were associated with development of right-sided heart failure during either the active or convalescent phase of infection than episodes in which smaller vegetations were noted (p = 0.02). No other grouping of clinical complications was significantly associated with vegetations larger than 1.0 cm.
Mitral Valve Prolapse To determine that the high prevalence of mitral valve prolapse was a "true-positive" and related to structural abnormalities of the mitral valve apparatus, rather than associated with disproportionately small ven tricles (Uvalvulo-ven tricular dispropor tion?"), values for LVESV and LVEDV were compared in 250
Table 3-Left Ventricular Function by Two-DimtmBional Echocardiography in Patients with Tricuspid Endocarditia with and without Mitral Valve Prolapae Mitral Valvular Prolapse Parameter
Present
Absent*
p Valuet
Increased LVEDV (>90 ml) Increased LVESV (>35 ml) Decreased LVEF «50 percent)
17/2:3 is/2:3 4/23
12126
0.05 <0.02 NS
8/26 5/26
*Four additional patients without mitral valvular prolapse had suboptimal studies for LVED~ LVES~ and LVEF and are not included in this analysis. tNS, Not significant.
patients with and without prolapse. Of interest, there
was a significant association between increases in LVESV and LVEDV and the presence of mitral valve
prolapse on the two-dimensional echocardiogram (Table 3); for example, 15 (65 percent) of23 episodes with mitral valve prolapse had increases in LVES~ as compared to only eight (31percent) of 26 episodes without mitral valve prolapse (p<0.02). Similarly, 17 (74 percent) of 23 episodes with, compared with 12 (46 percent) of 26 episodes without, mitral valve prolapse were associated with increased LVEDV (p=0.05). Moreover, values for LVEF in patients with and without mitral valve prolapse were similar. These data suggested that the mitral valve prolapse seen in patients with tricuspid endocarditis was caused by structural abnormalities of the mitral valve apparatus (rather than disproportionately small left ventricular size) and was associated with clinically silent mitral regurgitation. DISCUSSION
Over the past 15 years, the echocardiogram has developed into a valuable noninvasive tool for the diagnosis and management of patients with infective endocarditis. 20 The role of this technique in evaluating patients with infective endocarditis has been expanded even further since inception of the two-dimensional scanning techniques. Whereas the earlier M-mode scanners were limited to visualization of the mitral valve and only two aortic, one pulmonic, and one tricuspid leaflet, the two-dimensional scanning technique provided more complete delineation of all valvular leaflets. 21 Moreover; the size, shape, mobility, and points of attachment of vegetations are also better seen on two-dimensional echocardiographic studies than with the M-mode technique. In addition, vegetations on previously damaged valves are more easily identified by the two-dimensional scanner. A number of investigators have associated both the echocardiographic detection and size of vegetations on aortic and mitral valves with an increased risk of complicaTrIcuspid Valve Endocarditis due to S 8ureus (Bayerat 8/)
tions, particularly major peripheral emboli, congestive heart failure, frank valve disruption, and need for valvular replacement surgery. 7,22-25 In contrast, in patients with right-sided infective endocarditis, the prognostic implications of the size of the vegetation and detection on the two-dimensional echocardiogram have not been well delineated. Robbins et al" have recently reported a series of21 patients with right-sided valvular infective endocarditis and demonstrated that patients with large vegetations on the two-dimensional echocardiogram had a significantly worse prognosis with medical therapy. Thus, four (36 percent) of their 11 patients with vegetations greater than 1.0 em required surgery for prolonged pyrexia while receiving appropriate antibiotic therapy, as compared to none of eight patients with smaller vegetations. Some interpretive problems exist with this study. It was not clear whether the clinical evaluations were performed prospectively or retrospectively by chart review, and whether these were done by the same investigative team at the two institutions. Importantly, only 11 (48 percent) of the 23 episodes were caused by S aureus alone, with the remaining episodes being polymicrobial, caused by other aerobic Gram-positive cocci, or "culture-negative." Lastly, four (17 percent) of the 23 episodes, including one of the cases requiring tricuspid valve surgery, involved either the pulmonic valve alone (three cases) or the mitral valve in addition to the tricuspid valve. The multi-institutional aspects of this study, as well as the bacteriologic and valve-site diversities in their patients make interpretation of their observations problematic. Wong et al? also attempted to correlate larger tricuspid vegetations on M-mode or two-dimensional echocardiograms with a less favorable clinical outcome; however, only seven of their ten cases of tricuspid endocarditis were caused by S aureus; of note, two of the three patients who required tricuspid surgery had endocarditis caused by organisms other than S aureus (5 epidermidis and Pseudomonas aeruginosa in one case each). We designed the current study to include only patients with tricuspid endocarditis due to S aureus in order to avoid the interpretive difficulties associated with bacteriologic and valve-site diversities. Several interesting observations resulted from our investigation. Most patients (50/53; 94 percent) in our study were bacteriologically cured of active tricuspid endocarditis with antimicrobial therapy alone. Patients with tricuspid vegetations noted on two-dimensional echocardiograms remained febrile significantly longer despite appropriate therapy (about two weeks, mean) as compared to patients without demonstrable vegetations (about one week, mean); however, neither the actual size of the vegetation nor the presence of
vegetations larger than 1.0 cm was significantly correlated with the duration offever. Also, in contrast to leftsided infection, where the echocardiographic demonstration ofvegetations is associated with increased risks of major embolic episodes," this study of tricuspid endocarditis could not document such an association vis-a-vis recurrent pulmonary emboli while on appropriate therapy. Clinically evident right-sided heart failure was a relatively uncommon consequence of tricuspid endocarditis due to S aureus in our population, occurring in less than 20 percent of the episodes in either the active or convalescent phases of infection; however, the presence of tricuspid vegetations was significantly correlated with increases in RVED dimensions on initial two-dimensional echocardiograms. This finding suggests that patients with tricuspid endocarditis due to S aureus and vegetations on the two-dimensional echocardiogram are at Significantly increased risk for incipient right-sided heart failure. This presumption is further supported in our study by the fact that Significantly more patients with increased RVED dimensions (20 percent) eventually required tricuspid surgery than their counterparts with normal RVED dimensions (zero). The presence of vegetations larger than 1.0 em on two-dimensional echocardiograms in our patients with tricuspid endocarditis was not Significantly associated with any individual prognostic parameter on univariate analyses; however, the presence of such larger vegetations did significantly correlate with an increased propensity for such patients to develop clinical evidence of right-sided heart failure during either the active or convalescent phase of their infection. One must interpret this correlation with some caution, as the long-term followup in our study was not optimal. Only about onethird of the patients were followed for more than one year after discharge from the hospital. It is thus entirely possible that the frequency of developing clinically evident right-sided heart failure following tricuspid endocarditis due to 5 aureus is higher than we documented (about 20 percent). The high incidence of echocardiographically defined mitral valve prolapse (23/53; 43 percent) in this study was unexpected. No episodes were associated with clinical evidence of either the "click-murmur" syndrome" or mitral valve endocarditis on initial or followup evaluation. It is known that the prevalence of mitral valve prolapse on the two-dimensional echocardiogram is not different in uninfected drug addicts as compared with normal patients. Siegel et al26 from our own institution demonstrated that 6 percent of addicts actively using intravenous narcotic drugs, but without evidence of endocarditis, had mitral valve prolapse on two-dimensional echocardiograms. The echocardiographic demonstration of mitral valve prolapse appeared to represent structural alterations of the mitral CHEST I 93 I 2 I FEBRUARY, 1988
251
valve apparatus in our patients, rather than possible "false-positive" valve prolapse related to small ventricular dimensions. This was supported by finding significant increases in LVESV and LVEDV in patients with (vs without) mitral valve prolapse in our study, as well as also observing essentially normal LVEF in this population. Why our predominantly addicted population with tricuspid endocarditis due to S aureus has such a high prevalence of underlying mitral valve prolapse by two-dimensional echocardiograms remains unclear. One possibility is that patients with mitral valve prolapse also have a more generalized myxomatous degeneration of other valves, including the tricuspid valve. This has, in fact, been demonstrated in postmortem studies of patients with mitral valve prolapse. 27 It is also well documented that addicts are exposed frequently to S aureus bacteremia from intravenous narcotic self-injections, performed under suboptimal hygienic conditions, through cutaneous sites which are heavily colonized with staphylococci.28-30 Thus, it seems reasonable that the high prevalence of mitral valve prolapse in our study represents a "selection" of staphylococcemic patients with more global myxomatous degeneration of the atrioventricular valves predisposing to tricuspid endocarditis; however, this theory remains speculative. In summary, the demonstration of tricuspid vegetations on two-dimensional echocardiograms in patients with right-sided endocarditis due to S aureus identifies a subset of patients that is (1) likely to exhibit a longer time to defervescence with appropriate medical therapy, (2) likely to have findings of subclinical right-sided heart failure, and (3) at increased risk for eventual tricuspid surgery. Patients with vegetations larger than 1.0 em appear at increased risk for development of clinically evident right-sided heart failure during the active or convalescent phases of their infection. Our analysis confirms that the vast majority of patients with tricuspid endocarditis due to S aureus are cured medically without serious short-term or long-term sequelae, regardless of the presence of tricuspid vegetations on the two-dimensional echocardiogram, and that this latter finding alone should not dictate early empiric valvular surgery; however, the demonstration of tricuspid vegetations in this population does identify a subset of patients who require careful clinical and echocardiographic follow-up for possible long-term complications of endocarditis due to S aureus. A similar position has been recently stressed by Dinubile." ACKNOWLEDGMENTS: We thank Susan Partridge, R.N., for excellent technical assistance throughout this study.
REFERENCES 1 Abrams B, Sklaver A, Hoffman l: Greenman R. Single or combination therapy of staphylococcal endocarditis in intravenous drug abusers. Ann Intern Med 1979; 90:789-791 2 Menda KB, Gorbach SL. Favorable experience with bacterial
252
endocarditis in drug addicts. Am Heart J 1973; 78:25-32 3 Chambers HF, Korzeniowski OM, Sande MA. Staphylococcus aureus endocarditis: clinical manifestations in addicts and nonaddicts. Medicine 1983; 62:170-77 4 EI-Khatid MR, Wilson FM, Lerner AM. Characteristics of bacterial endocarditis in heroin addicts in Detroit. Am J Med Sci 1976; 271:197-201 5 Banks l: Fletcher R, Ali N. Infective endocarditis in heroin addicts. Am J Med 1973; 55:444-51 6 Robbins MJ, Frater RWM, Soeiro R, Frishman WH, Strom JA. Influence of vegetation size on clinical outcome of right-sided infective endocarditis. Am J Med 1986; 80:165-71 7 Wong 0, Chandraratna PAN, Wishnow RM, Dusitnanond ~ Nimalasuriya A. Clinical implications of large vegetations in infectious endocarditis. Arch Intern Med 1983; 143:1874-77 8 Baird-Parker AC. Gram-positive cocci. In: Buchanan RE, Gibbons NE, eds. Bergey's Manual of determinative bacteriology, 8th ed. Baltimore: Williams and Wilkins, 1974:478-528 9 Bayer AS, Lam K, Ginzton L, Norman DC, Chiu CY, Ward JI. Staphylococcus aureus bacteremia: clinical, serologic and echocardiographic findings in patients with and without endocarditis. Arch Intern Med 1987; 147:457-62 10 Bayer AS. Infective endocarditis with renal involvement. In: Glassock RJ, ed. Current therapy in nephrology and hypertension. Philadelphia: BC Decker, Inc, 1985:135-38 11 Ginzton LE, Siegel RJ, Criley JM. Natural history of tricuspid valve endocarditis: a two-dimensional echocardiographic study. Am J Cardiol 1982; 49:1853-59 12 Stratton JR, Werner JA, Pearlman AJ, Junko CJ, Kliman S, Jackson MC. Bacteremia and the heart: serial echocardiographic findings in 80 patients with documented or suspected bacteremia. Am J Med 1982; 73:851-58 13 Feigenbaum H. Echocardiography. Philadelphia: Lea and Febiger, 1986:636 14 Ginzton LE, Brazelton M, Delaney ~ Adomian G, Garner 0, Laks MM. Quantitation of two-dimensional echocardiography: day-to-day variability of left ventricular mass in the conscious dog. Clin Res 1983; 31:9A 15 Levinson ME. Infective endocarditis: response to therapy. In: Kaye D, ed. Infective endocarditis. Baltimore: University Park Press, 1976:185-89 16 Saravolatz LD, Pohlod OJ, Arking LM. Community-acquired methicillin-resistant Staphylococcusaureus infections. Ann Intern Med 1982; 97:325-29 17 Dixon WJ. BMDP statistical software. Berkeley, Calif: University of California Press, 1983 18 Arbulu A, Asfaw I. Tricuspid valvulectomy without prosthetic replacement: ten years of clinical experience. J Thorac Cardiovasc Surg 1981; 82:684-91 19 Ginzton LE. Prolapsed mitral leaflet syndrome. In: Shine KI, ed. New York: John Wiley and Sons, 1983:1-20 20 Popp RL. Echocardiography and infectious endocarditis. In: Remington JS, Swartz MN, eds. Current clinical topics in infectious diseases. New York: McGraw-Hill Book Co, 1983:98110 21 Mintz GS, Kotler MN. Clinical value and limitations of echocardiography: its use in the study of patients with infectious endocarditis. Arch Intern Med 1980; 140:1022-26 22 Stewart JA, Silimperi D, Harris ~ Wise NK, Fraker TD, Kisslo JA. Echocardiographic documentation of vegetative lesions in infectious endocarditis: clinical implications. Circulation 1980; 61:374-80 23 Mintz GS, Kotler MN, Segal BL, Parry W A comparison of twodimensional and M-mode echocardiography in the evaluation of patients with infective endocarditis. Am J Cardiol 1979; 43: 738-44 24 Wann LS, Hallam CC, Dillon jc, Weyman AE, Feigenbaum H. Tricuspid Valve Endocarditis due to S aureus(Bayer et aJ)
25
26 27 28
Comparison of M-mode and cross-sectional echocardiography in infective endocarditis. Circulation 1979; 60:728-733 Davis RS, Strom jA, Frishman WH, Becker R, Matsumoto M, LeJemtel TH, et ale The demonstration of vegetations by echocardiography in bacterial endocarditis: an indication for early surgical intervention. Am J Med 1980; 69:57-63 Siegel RJ, Guze RA, Beland K, Guze L, Criley JM. Twodimensional echoeardiographic findings in drug addicts. J Cardiovasc Ultrasonography 1982; 1:133-34 Pomerance A. Ballooning deformity (mucoid degeneration) of atrio-ventricular valves. Br Heart J 1969; 31:343-51 Crane LR, Levine D~ Zervos MJ, Cummings G. Bacteremia in
narcotic addicts at the Detroit Medical Center: microbiology, epidemiology, risk factors and empiric therapy. Rev Infect Dis
1986; 8:364-73
29 TuazonCU, Sheagren IN. Increased rate of carriage of Staphylococcus aurew among narcotic addicts. J Infect Dis 1974; 129: 725-27 30 TuazonCU, Sheagren IN. Staphylococcal endocarditis in parenteral drug abusers: source of the organism. Ann Intern Med 1975; 129:327-29 31 DiNubile M. Surgery for addiction-related tricuspid valve endocarditis: caveat emptor. Am J Moo 1987; 82:811-13
Current Concepts in Asthma, Allergy and Pulmonary Disease Beth Israel Medical Center and the Page and William Black Post-Graduate School of Medicine of the Mount Sinai School of Medicine, will present this program at Beth Israel Medical Center; April 29. For information, contact Beverley R. Baptiste, Medical Education, Beth Israel Medical Center; First Avenue at 16th Street, New York 10003 (212:420-2849).
Critical Care Medicine '88 The Orlando Regional Medical Center will present this program April 6-9 at the Buena Vista Palace Hotel, Lake Buena Vista, Florida. For information, contact Patti L. Devlin, CME Coordinator, Orlando Regional Medical Center, 1414 South Kuhl Avenue, Orlando 32806-2093 (305:841-5144).
CHEST I 93 121 FEBRUARY. 1988
253