The trusted source for
healthcare information and
Blood Culture-negative Endocarditis: What Can the Laboratory Bring to the Table?
Abstract & Commentary
By Ellen Jo Baron, Ph.D., D(ABMM), Professor Emerita, Stanford University School of Medicine, Director of Medical Affairs, Cepheid, Sunnyvale, CA, is Associate Editor for Infectious Disease Alert.
Dr. Baron reports no financial relationships relevant to this field of study.
Synopsis: The etiology of blood culture-negative endocarditis was identified in 62.7% of 759 patients using combinations of serological, molecular, and histopathological assays. The majority of the agents not detected by conventional cultures were Coxiella burnetii (30%) and Bartonella spp. (11%), both diagnosed using serological tests, and Trophyrema whipplei, other unusual bacteria, and fungi making up the remainder. For 14% of patients, diagnosis required PCR testing on valve tissue removed during surgery (not in paraffin). The cause of disease could not be determined for 264 patients; 2% of patients had non-infectious endocarditis.
Sources: Pierre-Edouard Fournier, et al. Comprehensive diagnostic strategy for blood culture negative endocarditis: A prospective study of 819 new cases. Clin Infect Dis. 2010;51:131-140.
Several years ago, two different publications appeared showing that prolonged incubation of blood cultures beyond the initial five-day protocol for patients with suspected fastidious bacterial endocarditis did not yield significant additional pathogens with today's modern blood-culture media and automated methods.1,2 Many laboratories, however, still have problems convincing clinicians that such measures won't be productive. The current publication by the renowned Raoult group confirms the findings of those less sophisticated studies, and suggests an algorithm that will generate the diagnosis in the majority of circumstances in which a diagnosis is possible. Unfortunately, almost no other laboratories have the capabilities of the Raoult laboratory, so that clinicians outside France should not expect to see the same results from their local, or even their reference, laboratory.
This group has pioneered the use of cutting edge molecular technologies for detecting pathogens in the bloodstream and in tissues, and has actively pursued development and documentation of new tools for diagnosis of this syndrome for more than 15 years.3-6 In my opinion, it is the most competent laboratory in the world to find an etiological agent of endocarditis when all else has failed. Unfortunately, laboratories in the United States are prevented from sending samples to France because of regulatory restrictions. Reference laboratories must be accredited by the state and CLIA, for example, in California, and laboratories accredited by the College of American Pathologists are supposed to send tests out only to other CAP-accredited laboratories. Neither regulatory body is recognized in France.
In this most recent publication, the group investigated the etiology of disease in 819 patients suspected of having blood culture-negative endocarditis (BCNE); approximately 213 involved native valves. Detailed clinical histories were obtained, along with several types of samples, when possible, including serum, EDTA and heparinized blood samples, excised valve tissue frozen at -80° C, and tissue in paraffin blocks, all received over the eight years of the study period. Sixty patients without a diagnosis of endocarditis, based on the modified Duke Criteria, were originally excluded.7
The agents tested for by serological testing were Coxiella, Bartonella, Legionella, and Mycoplasma pneumoniae. Blood, heart-valve tissue, and other tissues were extracted, and specific primers employed, for PCR testing after initial tests yielded no results. Primers for all bacteria, all fungi, and individual species Coxiella, Bartonella, Tropheryma, and Chlamydia, as well as cytomegalovirus and enteroviruses were used. Tissue cultures were set up on tissues, and if bacteria were seen by any of a number of fluorescent and histological stains, they were identified by universal primer amplification and sequencing. Tissues in paraffin block were stained with immunoperoxidase, acid-fast, and additional special stains such as Warthin-Starry and other silver stains. Autoimmunohistochemistry was used on tissues for which no etiology was detected by other means. A number of tests were employed to detect autoantibodies, such as rheumatoid factor, anti-DNA, and antinuclear antibodies.
Nineteen of the 759 patients included in the study group had non-infectious endocarditis, including marantic endocarditis (7), Libmann-Sacks endocarditis (4), and Behcet disease (1). Seven more patients had autoimmune disease diagnosed elsewhere by additional tests. By the Duke criteria, 549 of 740 remaining patients were designated as having definite endocarditis, and Raoult's group was able to detect an etiological agent in 476 (86.7%)of those. One-hundred ninety-one patients (~ 26%) were classified as "possible" endocarditis. Serologic tests provided the diagnosis in almost half of the patients, emphasizing the importance of this modality, often overlooked by clinicians in their zeal to order molecular assays.
Employing all possible tests, the authors detected 315 cases due to Q fever (Coxiella), 86 due to Bartonella, 12 due to T. whipplei, eight fungi, and 70 conventional bacteria, predominantly streptococci (many S. gallolyticus) and staphylococci. Readers of Infectious Diseases Alert may remember that S. gallolyticus was discussed in a previous Infectious Disease Alert article on the recent changes in the taxonomy of the S. bovis group.8 PCR on EDTA blood samples provided the diagnosis in < 15% of all patients. In fact, PCR tests using blood detected only three of the seronegative patients, although valve biopsies were more fruitful, providing a bacterial diagnoses for 66.1% of samples and a fungal diagnosis for 3% among patients from whom valve samples were available. Cell culture never revealed an agent that was not detected by another method, for which microbiology laboratories can breathe a sigh of relief. Valve biopsies were positive by culture more often than blood as well, yielding an organism in 58 of 127 patients (46%). PCR of valves yielded a positive result in 157 of 227 cases (69%).
The authors summarized their findings with a proposed algorithm. It must be remembered that the majority of patients lived in France, other European countries, and Algeria, and only a few patients from the United States were studied, so the results of a similar series might be expected to have different outcomes if conducted in the United States. The authors of this series suggest that after the initial negative blood cultures, serum should be the first specimen to evaluate for Q fever and Bartonella, followed by specific PCR assays for Bartonella species, T. whipplei, and fungi. For patients with negative results at this point, additional serological and PCR tests could be employed. They also suggested that tests for antinuclear antibodies and rheumatoid factor should be instituted at this point. The authors also recommend that heart valve biopsies should first be tested using broad-range PCR assays for bacteria and fungi, with autoimmuno-histochemistry only for patients with all other results negative. Although they had tested the commercial Roche SeptiFast blood culture multi-analyte PCR system, available in Europe but not FDA-cleared in the United States, they found its sensitivity to be so sufficiently poor that they did not recommend its use. Instead, they suggested that if laboratories had specific staphylococcal and streptococcal PCR assays available, they should be used. Of course, right now in the United States there are no commercial tests available for use on blood directly from the patient without the benefit of culture amplification in blood cultures.
In summary, serological tests are the second most productive modality for diagnosis of culture-negative endocarditis, followed by PCR, and then even more heroic measures, if they are available. Even then, one-third of cases remained for which an etiology could not be determined.