Bacteriological analysis of cardiac valves might be indicated in patients with suspected endocarditis.
We report here a prospective study on fifty-three consecutive patients whose native valves were sent to the bacteriological and pathological laboratories, to investigate the performance of direct microscopic examination of imprints and valve culture.
On the basis of a histopathological gold standard to classify the inflammatory valve process, the sensitivity, the specificity, the positive and the negative predictive values of direct microscopic examination of imprints and valve culture were 21%, 100%, 100%, 60%, and 21%, 72%, 38%, 52% respectively. This weak threshold of the direct microscopic examination of imprints could be due to antimicrobial therapy prescribed before cardiac surgery and the fact that the patients came from a tertiary hospital receiving patients with a prolonged history of endocarditis.
Clinical context and histopathology are indispensable when analyzing the imprints and valve culture.
Infective endocarditis accounts for a small fraction of the total number of replaced valves. Histopathologic examination of these tissues removed during surgery is a standard practice in most hospitals. Direct microscopic examination of imprints of the valve is not usually requested. However, bacteriological analysis of cardiac valves might be advised in patients with suspected endocarditis . Laboratories often perform analysis without clinical information. We report here a prospective study on 53 native valves sent to both bacteriological and pathological laboratories to investigate the performance of direct microscopic examination of imprints (DMEI) and valve culture to confirm and/or find the etiologic agent of acute endocarditis [2,3]. The aim of the study was to determine if routine direct microscopic examination of imprints and the culture of the native valves are informative for confirming and diagnosing endocarditis, using the histopathology as gold standard defining an evolutive inflammatory process.
Materials and methods
53 native valves were removed aseptically from patients submitted to cardiac surgery. They were sampled and carried out in sterile containers without medium. The heart valves were manipulated under a laminar flow unit. Portions of the valve showing disease (vegetations, thrombi) were separated into two parts: the first for bacteriological study and the second for pathological study. Before adding formalin, smears of the lesions were made on sterile slides before staining (Gram, May-Grünwald-Giemsa, Gimenez stains) for direct microscopic examination of imprints.
The valve tissue was ground in 1 ml of brain-heart broth with a mortar and pestle, and then cultured onto 5% sheep blood, chocolate + isovitalex agar and in a paraffined brain-heart broth and Todd-Hewitt broth. The plates were incubated both in aerobic atmosphere and under carbon dioxide atmosphere (5%) for 10 days at 37°C. The broths were kept for 1 month (with subcultures at day 8 and day 30) [2,3]. A chocolate agar was incubated under carbon dioxide atmosphere (5%) for 1 month for Bartonella sp detection. A piece of the valve was deep frozen at -80°C. Standard microbiological techniques were performed to identify the bacteria. Results of the May-Grünwald-Giemsa and Gram stains were reported as direct microscopic examination of imprints as follows: rare (≤ 1 per oil immersion [X 1000] field), 1+ (2–9 per oil immersion [X 1000] field), 2+ (10–50 per oil immersion [X 1000] field), 3+ (> 50 per oil immersion [X 1000] field).
The tissue specimens were fixed in 10% neutral formalin and embedded in paraffin. Consecutive 5 μm sections were stained by hematoxylin-eosin-saffron-astra blue (HES), Gram, Gomori and Periodic Acid Shiff stain methods. The HES allowed us to classify the inflammatory response of the host as acute or chronic/sequellar inflammation  considered as the gold standard for evaluating the bacteriological results. The other stains gave information about the presence or absence of microorganisms: this step was considered a direct pathological examination (DPE).
53 patients admitted to the cardiology hospital were included in this study over two years (February 1997–February 1999). After the valve laboratory studies, clinical features and other biological parameters were retrospectively collected as follows: blood cultures performed before and during surgery, antimicrobial therapy prescribed, intracellular bacteria serologies (Bartonella, Coxiella, Chlamydiae, Legionella) if the data were available. The number of positive blood cultures was not always available in the results obtained before surgery.
Results and discussion
During this prospective study over 2 years, 989 native cardiac valves were removed in the cardiology ward. Among these, 53 (5.3%) were sent to both pathology and bacteriology laboratories because endocarditis was suspected during surgery or because the patient was known to have suffered an endocarditis in the previous months. These 53 valves came from 42 men and 11 women (mean age 50.4 years). The valves were classified as follows: 19 mitral valves, 31 aortic valves, 2 tricuspid valves and 1 pulmonary valve. The pathological classification revealed 24 evolutive acute infectious processes considered as true positive and 29 chronic/sequellar inflammatory processes considered as true negative. The detailed results for each patient are reported in Tables 1 (evolutive acute infections) and 2 (chronic/sequellar inflammatory process). Among the 24 with evolutive acute endocarditis, 23 patients were classified with a definite infective endocarditis according to Duke University : two clinical major criteria +/- one pathologic criteria. Among the 29 other patients, 9 were classified with a definite infective endocarditis according to Duke University and 20 as possible infective On the basis of the histopathological gold standard, the sensitivity, the specificity, the positive predictive value (PPV) and the negative predictive value (NPV) of the DMEI and of the culture are reported on Table 3. The performance of the DPE is also reported in the Table 3. The weak sensitivity of the DMEI could be explained by the thickness of the tissue in contact on the smears performed, and by the antimicrobial therapy decreasing the inoculum. The DPE performed in the tissue is more interesting, giving a sensitivity of 71 % and a specificity of 90%. Similar results were described by Chuard et al.  revealing the underfullness of direct examination because a negative result does not eliminate the diagnosis in treated patients. The status of the histopathological lesion seems to be more informative of the evolution of the illness. This could be due to the antimicrobial therapy prescribed before cardiac surgery. In our study, 78% of patients had received antibiotics at the time of cardiac surgery. Prescribed antibiotics were always beta-lactams with or without another antibiotic. If DMEI was positive (patients 2, 4, 5, 8, 9), that did not necessarily indicate antibiotic failure because the bacteria observed were dead.
Table 1. Evolutive acute infections: results of DMEI, DPE, culture of valve and clinical features and other biological parameters
Table 2. Chronic/sequellar processes: results of DMEI, DPE, culture of valve and clinical features and other biological parameters
Table 3. Performance of the DMEI, DPE and culture of cardiac native valves
The valve cultures were negative in 40 patients. In the 33 patients with retrospective documentation, 84 % (28/33) received antibiotics and 60 % (20/33) had positive blood cultures before cardiac surgery. In those with acute endocarditis, 71 % (15/21) had positive blood cultures. The intracellular bacteria serologies (Bartonella, Coxiella, Chlamydiae, Legionella) were only performed in 3 patients (patients 6, 19, 26) and were all negative. Patients 2 and 4, with a positive DMEI, were negative for valve culture. The antimicrobial therapy prescribed 4 weeks before surgery corresponding to the last positive blood culture, had probably sterilized the lesions. Valve culture was positive in 13 patients. 12 were evaluable and corresponded to 4 acute inflammatory processes and 8 chronic/sequellar processes. For the 8 true negative patients, except for patients 28 and 29 where bacterial blood cultures were known before surgery, the bacteria found in the valve could probably have been a contamination since the patients did not present evolutive clinical symptoms of endocarditis. As shown by Giladi et al., valve culture is not recommended when acute endocarditis is not suspected. In the 4 endocarditis patients, 2 (patients 1 and 15) with no antimicrobial treatment had positive valve culture corresponding to the etiologic agent (Staphylococcus epidermidis). As described since 1990, the coagulase-negative staphylococci have been more and more implicated in native endocarditis . The 2 other valve cultures were difficult to evaluate. Patient 20 was known to have had a blood culture documented on with a classical etiologic agent (Streptococcus viridans) and the mixed valve culture was probably a contamination. In another patient (patient 19) had a positive DMEI (Gram positive rod), a positive valve culture with Propionibacterium acnes, but a blood culture yielding Haemophilus influenzae. The microbiologist could not conclude anything from these results. P. acnes may often be considered as a contaminant, but in a few cases  this bacterium has been responsible for true endocarditis on native valves. This weak threshold of the direct microscopic examination of imprints could be due to antimicrobial therapy prescribed before cardiac surgery and the fact that the patients came from a tertiary hospital receiving patients with a prolonged history of endocarditis.
Although DMEI is a major criterium of the von Reyn classification and is often rapidly available, it was found to be poorly informative in the assessment of endocarditis in this study. The presence of acute inflammation is the crucial point to evaluate the status of the endocarditis process. If in an acute context, no etiologic agent is known, such as fastidious bacteria , it is important to perform valve culture. The drawback of this technique is prior antibiotic therapy giving a poor yield. If the valve has undergone a chronic/sequellar process, the culture will not be informative. In this situation or in acute endocarditis, in a patient with no bacterial documentation and/or receiving antimicrobial treatment, it might be interesting to introduce routine molecular techniques . Clinical context and histopathology are essential to validate the results of valve culture .
Arch Pathol Lab Med 1998, 122:412-415. PubMed Abstract
Mainardi JL, Vandenesch F, Casalta JP, N'Guyen J, Benoît C, Tissot-Dupont H, Maugein J, Weber M, Bouvet A, Etienne J, et al.: Recommandations pour le diagnostic microbiologique et l'étude anatomopathologique des valves cardiaques au cours des endocardites infectieuses.
Ann Pathol 1995, 15:221-223. PubMed Abstract
Am J Med 1994, 96:200-209. PubMed Abstract
Giladi M, Szold O, Elami A, Bruckner D, Johnson BL: Microbiological cultures of heart valves and valve tags are not valuable for patients without infective endocarditis who are undergoing valve replacement.
Clin Infect Dis 1997, 24:884-888. PubMed Abstract
Br Heart J 1990, 64:381-384. PubMed Abstract
J Clin Microbiol 1994, 32:3043-3045. PubMed Abstract
Peterson LR, Hamilton JD, Baron EJ, Tompkins LS, Miller JM, Wilfert CM, Tenover FC, Thomson RB Jr: Role of clinical microbiology laboratories in the management and control of infectious diseases and delivery of health care.
The pre-publication history for this paper can be accessed here: