Professional Documents
Culture Documents
295–304 (2009)
Lipopolysaccharide-Binding Protein:
A New Biomarker for
Infectious Endocarditis?
Tanja Vollmer,1 Cornelia Piper,2 Knut Kleesiek,1 and Jens Dreier1*
BACKGROUND: Infectious endocarditis (IE) is a bacterial Infectious endocarditis (IE),3 a relatively rare disease
infection of the endocardium. Diagnosis is based on with an incidence of 30 per million population (1 ), is
results obtained from echocardiography, blood cul- associated with high rates of mortality and morbidity.
tures, and molecular genetic screening for bacteria and Rapid diagnosis of IE is critical for effective treatment.
on data for inflammatory markers such as the leuko- Diagnosis is based primarily on the results of echocar-
cyte (WBC) count and the C-reactive protein (CRP) diographic examinations and blood culture tests [Duke
concentration. The aim of the present study was to criteria (2 )], as well as on the results of serology tests
evaluate lipopolysaccharide-binding protein (LBP) as a and molecular genetic screening methods for bacteria
supportive biomarker for the diagnosis and therapeu- in cases of heart valve replacement (3, 4 ). The initial
tic monitoring of IE. diagnosis of IE can be difficult, however, because of
nonspecific clinical symptoms, negative results in
METHODS: We measured LBP and CRP concentrations blood cultures, and ambiguous echocardiographic
and WBC counts in 57 IE patients at hospital admis- findings. In this context, the availability of biomarkers
sion, 40 patients with noninfectious heart valve dis- is of major interest. Several studies have investigated
eases (HVDs), and 55 healthy blood donors. The pro- the application and usefulness of important laboratory
gression of these 3 markers and the influence of cardiac markers for IE diagnosis (5–12 ). Other studies (13–16 )
surgery on them were evaluated in 29 IE patients and 21 have focused on the clinical usefulness of serum C-
control patients. reactive protein (CRP), a marker most frequently used
in Europe (13 ), in IE. CRP has been assumed to be
RESULTS: Serum LBP concentrations were significantly primarily an indicator of local bacterial infections, but
higher in IE patients [mean (SD), 33.41 (32.10) mg/L] increased CRP concentrations have also been strongly
compared with HVD patients [6.67 (1.82) mg/L, P ⬍ associated with various nonbacterial infectious dis-
0.0001] and healthy control individuals [5.61 (1.20) eases, e.g., rheumatic disorders (17 ). CRP has been
mg/L]. The progression in the LBP concentration dur- shown to have a limited ability to distinguish between
ing therapy of IE patients correlated with the changes in bacterial and viral infections (18 ).
the CRP concentration. The 2 markers were equally Lipopolysaccharide-binding protein (LBP), a gly-
influenced by antibiotic treatment and surgical cosylated 58-kDa protein produced predominantly in
intervention. hepatocytes, has recently been described as a promising
novel diagnostic marker for the diagnosis of local bac-
terial infection (19 –23 ). LBP is released constitutively
CONCLUSIONS: Serial LBP measurement may provide an
into the bloodstream upon acute-phase stimulation
effective and useful tool for evaluating the response to (24 ). It has a high affinity for lipopolysaccharide (LPS)
therapy in IE patients. We found a strong correlation and other bacterial-surface components (25 ). Binding
between LBP and CRP concentrations; LBP has a ten- of LPS to LBP facilitates the transportation of endo-
dency to increase earlier in cases of reinfection. toxin molecules to immune effector cells possessing
© 2008 American Association for Clinical Chemistry
surface CD14 receptors and further activates the proin-
flammatory cascade (24, 26 ). Association of the LPS/
1
Institut für Laboratoriums- und Transfusionsmedizin, Herz- und Diabeteszen- sitätsklinik der Ruhr-Universität Bochum, Georgstrasse 11, 32545 Bad Oeyn-
trum Nordrhein-Westfalen, Universitätsklinik der Ruhr-Universität Bochum, Bad hausen, Germany. Fax ⫹49-5731-97-2307; e-mail jdreier@hdz-nrw.de.
Oeynhausen, Germany; 2 Klinik für Kardiologie, Herz- und Diabeteszentrum Received March 4, 2008; accepted August 26, 2008.
Nordrhein-Westfalen, Universitätsklinik der Ruhr-Universität Bochum, Bad Previously published online at DOI: 10.1373/clinchem.2008.106195
3
Oeynhausen, Germany. Nonstandard abbreviations: IE, infectious endocarditis; CRP, C-reactive protein;
* Address correspondence to this author at: Institut für Laboratoriums- und LBP, lipopolysaccharide-binding protein; LPS, lipopolysaccharide; HVD, nonin-
Transfusionsmedizin, Herz- und Diabeteszentrum Nordrhein-Westfalen, Univer- fectious heart valve disease; WBC, leukocyte; CI, confidence interval.
295
LBP complex with HDL inactivates LPS, thereby pro- cording to the manufacturer’s instructions. All samples
tecting against LPS toxicity (24 ). were tested in random order with controls and calibra-
The aims of this study were to evaluate (a) the tion curves incorporated into each run. Blood samples
clinical usefulness of LBP as a supportive biomarker for were collected in EDTA Monovettes (Sarstedt), and
IE diagnosis, (b) the applicability of LBP for monitor- WBCs were counted directly with the Cell-Dyn 3500
ing the response to antibiotic therapy, and (c) the in- (Abbott Diagnostics).
fluence of cardiac surgery on the progression of LBP
values in both IE patients and patients with noninfec- STATISTICAL ANALYSIS
tious heart valve disease (HVD). The use of LBP as a All data are presented as the mean (SD). The Mann–
marker was compared with the frequently used endo- Whitney U-test and the Student t-test were used for
carditis markers CRP and the leukocyte (WBC) count. statistical analyses as appropriate, and the assumption
of a gaussian distribution was tested in all data sets with
Materials and Methods the Kolmogorov–Smirnov test. We used Pearson and
Spearman correlation coefficients to assess correlations
PATIENTS AND CONTROLS between variables. P values ⬍0.05 were considered sta-
Between May 2005 and November 2007, serum sam- tistically significant. GraphPad Prism 4.0 software
ples were obtained from 57 patients with definite IE (GraphPad Software) was used for statistical analyses.
diagnosed according to the Duke criteria. The caus-
ative pathogens were identified with standard blood- Results
culture methods or broad-range PCR of the gene for
bacterial 16S ribosomal DNA, as described previously Table 1 summarizes the clinical characteristics of the
(27 ). Patients with immunosuppression, cardiogenic patients and the control cohorts. LBP was measured in
shock, or septic shock were excluded. Three of the pa- serum samples obtained from 57 patients with 60 epi-
tients had experienced 2 independent episodes of IE. sodes of IE, a control group of 40 HVD patients, and a
Patients were subdivided into 4 groups according to second control group of 55 healthy blood donors. We
the duration of antibiotic treatment [0 –10 days (n ⫽ confirmed the manufacturer’s LBP reference interval
38), 11–20 days (n ⫽ 7), 21–30 days (n ⫽ 6), and ⬎30 in our healthy control population. Fig. 1A presents
days (n ⫽ 9)]. Serum samples were also collected from scatter plots of LBP concentration for the 3 cohorts.
40 control HVD patients before surgical intervention. Serum LBP concentrations were increased in 58 epi-
HVDs consisted of noninfectious and nonrheumatic sodes of IE, and the increases were independent of the
valve insufficiencies or stenosis. We measured serum duration of antibiotic treatment. In 2 episodes, the LBP
LBP values in 55 healthy blood donors to confirm the concentration was within the reference interval. LBP
previously determined reference interval of the com- concentrations were significantly higher in IE patients
mercially available LBP assay. than in the 2 control groups [IE patients, 33.41 (32.10)
To examine the progression in LBP and CRP con- mg/L; HVD patients, 6.67 (1.82) mg/L; healthy blood
centrations and the WBC count, we collected serum donor control group, 5.61 (1.20) mg/L] (see Fig. 1A).
samples and EDTA-anticoagulated blood samples The diagnostic sensitivity was 96.7% [95% confidence
from 29 IE patients and 21 control patients with HVD interval (CI), 92.2%–100.0%], and the diagnostic
during their admission at our hospital. Samples were specificity was 87.5% (95% CI, 77.3%–97.7%) with re-
collected daily. The study protocol was approved by the spect to the HVD patients and 94.5% (95% CI, 88.5%–
institutional review board, and all patients gave their 100.0%) with respect to the healthy controls. The mean
informed consent. LBP concentration of the IE patient samples was 83.2%
higher than the mean concentrations of the control
MEASUREMENT OF CRP, LBP, AND THE WBC COUNT groups.
Serum samples were collected in serum Monovettes With respect to the duration of antibiotic treat-
(Sarstedt). The samples were clotted, centrifuged at ment, LBP concentrations were considerably lower af-
4000g for 10 min, and stored at ⫺70 °C within 2 days of ter 11–20 days and ⬎30 days of treatment [14.56 (5.29)
collection. Samples were stored at this temperature un- mg/L and 15.71 (8.05) mg/L, respectively], compared
til assayed. with treatment for 0 –10 days [41.97 (37.20) mg/L]
Serum CRP was measured with the ultrasensitive (Fig. 1A, inset). Comparatively high LBP concentra-
latex immunoassay CRP Vario (Abbott Diagnostics), tions were observed after 21–30 days of antibiotic treat-
and the total LBP concentration in serum samples was ment [25.45 (7.30) mg/L]. The serum concentration of
measured with a commercially available chemilumi- CRP in patients upon hospital admission showed a
nescence assay (IMMULITE LBP, Diagnostics Prod- trend similar to that of LBP, with IE patients having
ucts Corporation/Siemens Healthcare Diagnostics) ac- significantly higher concentrations than the individu-
Fig. 2. Different progressions of LBP and CRP concentrations and the WBC count during 3 IE episodes.
Examples of the progression of LBP and CRP concentrations and the WBC count during an uncomplicated IE episode (A), a
noticeably low LBP response during an IE episode (B), and a complicated IE course with recurrent infectious episodes (C). Arrows
indicate the onset of the LBP concentration increase compared with the CRP concentration. X, initiation of antibiotic treatment;
S, surgical intervention.
plicated IE course. The patient experienced 28 days of LBP concentrations in IE or the potential of LBP as a
recurrent infectious episodes while receiving antibiotic diagnostic marker for this disease.
treatment. LBP and CRP concentrations showed re- In this study, we evaluated the clinical applicability
lated progression patterns; however, postoperative in- of LBP as a supportive biomarker for IE diagnosis and
creases in LBP values occurred 3 days earlier than the monitoring disease progression and compared the re-
increases in CRP concentrations. A similar progression sults with those obtained for CRP and the WBC count.
pattern was observed in 2 other cases with recurrent LBP and CRP values were significantly higher in IE
infectious episodes (data not shown). WBC counts patients upon their admission to the hospital than in
were increased considerably compared with an un- the HVD patients and the healthy controls (Fig. 1, A
complicated IE course. and B). Mean LBP concentrations decreased with pre-
The progression in LBP and CRP concentrations vious antibiotic treatment, with a notable decrease ap-
and the WBC count was measured before and within 5 parent after 10 days of antibiotic treatment, although
days after surgical intervention in 29 IE patients and 21 comparatively high concentrations were still observed
HVD patients. At the time of surgery, LBP and CRP in some patients after 21–30 days of antibiotic treat-
concentrations were within the reference interval in 5 ment. The patients with the higher LBP concentrations
IE patients and 2 IE patients, respectively, and the WBC during this later time interval experienced IE episodes
count was within the reference interval in 14 IE pa- with major complications and severe bacteremia. This
tients. LBP and CRP concentrations and the WBC observation leads to the assumption that the occur-
count were within their respective reference intervals rence of high LBP concentrations after prolonged an-
for all HVD patients. Preoperative LBP and CRP con- tibiotic treatment may be correlated with a compli-
centrations remained significantly higher in IE patients cated disease progression; however, because our study
than in HVD patients (P ⬍ 0.0001; Fig. 3, A and B). cohort included only 15 patients with long-term anti-
biotic treatment (⬎20 days), this hypothesis will have
Similarly, the preoperative WBC count was higher in IE
to be validated with a larger group of patients to ex-
patients than in HVD patients (P ⬍ 0.001, Fig. 3C);
clude the possibility that these observations are merely
however, more than half of the evaluated IE patients
random outliers. We did not observe a statistically sig-
had WBC counts within the upper portion of the ref-
nificant difference in LBP values linked to the causative
erence interval. Peak LBP concentrations were ob-
pathogen (data not shown); however, the pathogen
served 24 – 48 h after surgical intervention [34.64
species and the access of antibiotics to the infected area
(12.85) mg/L in IE patients vs 33.90 (10.12) mg/L in
of the heart valve still influence the effectiveness of
HVD patients, Fig. 3A]. Peak CRP concentrations were antibiotic treatment and thus the reduction in the in-
also observed during this period [179.40 (59.84) mg/L flammatory trigger. We are aware of only one other
in IE patients vs 177.40 (50.82) mg/L in HVD patients, published study regarding the value of serum LBP
Fig. 3B]. In contrast, peak WBC counts were observed measurements at hospital admission for the diagnosis
0 –12 h after surgical intervention. We no longer ob- of IE, and the results of that study contradict those of
served a statistically significant difference between IE our study (12 ). Watkin et al. (12 ) observed that se-
patients and HVD patients in LBP values, whereas rum LBP concentrations were slightly but not signifi-
these 2 patient groups showed statistically significant cantly higher in IE patients than in controls. The
differences in the CRP concentration and the WBC mean LBP concentration was noticeably higher in the
count up to 12 h after surgical intervention (see Fig. 3, Watkin et al. control group, with a broader range
B and C). (mean, 15.76 mg/L; range, 8.97–30.73 mg/L), than in
our control group (mean, 5.62 mg/L; range, 3.50 –
Discussion 8.90 mg/L). One possible explanation for these diver-
gent results is that the study of Watkin et al. and our
Several studies have reported increased LBP serum study used different LBP assays.
concentrations in patients with severe sepsis due to We found the diagnostic sensitivities and specific-
gram-positive, gram-negative, and fungal infections ities of both LBP and CRP to be high (⬎90%). The
(19, 20 ). IE is a bacterial infection comparable to sep- WBC count had a low diagnostic sensitivity (26.7%)
sis, but infection of the endocardium usually is associ- but a high diagnostic specificity (100%). All of the ex-
ated with a low-level and noncontinuous bacteremia, a amined markers had nearly equivalent high positive
situation that contrasts with that of sepsis. To date, the predictive values. The WBC count had the lowest neg-
CRP concentration and the WBC count have com- ative predictive value (67.9%), whereas LBP and CRP
monly been used for IE diagnosis and monitoring a had equivalent high negative predictive values (97.8%
patient’s response to therapy. No systematic data have and 96.8%, respectively). In short, serum LBP and CRP
previously been published on the progression of serum concentrations were the most sensitive laboratory vari-
influences on LBP caused by HVDs or LBP fluctuations use of LBP as a marker instead of CRP. The value of
in healthy individuals; however, the typical clinical sit- LBP as a marker for early recognition of a new local
uation in the diagnosis of IE is a patient with prolonged septic crisis that we have highlighted has to be thor-
fever and other nonspecific complaints. In a follow-up oughly established in further investigations. Neverthe-
clinical study, a cohort of other patients with fever less, we propose the use of LBP as a marker in addition
would be an ideal control group and would provide to CRP for the diagnosis of IE and for monitoring the
additional data regarding the sensitivity and specificity response to therapy, particularly in cases with recur-
of LBP for evaluating fever. Also beneficial would be rent infection.
adequate patient cohorts for evaluating LBP for distin-
guishing between bacterial infection, viral infection,
and general inflammatory responses. Another general
problem is the heterogeneity of the IE patient cohort. Author Contributions: All authors confirmed they have contributed to
Patients exhibit different infection severities, show in- the intellectual content of this paper and have met the following 3 re-
dividual inflammatory responses, and may be at differ- quirements: (a) significant contributions to the conception and design,
ent stages of antibiotic treatment. Hence, test results acquisition of data, or analysis and interpretation of data; (b) drafting
primarily depend on the patient population and pretest or revising the article for intellectual content; and (c) final approval of
the published article.
probabilities. The advantages of LBP in cases of rein-
fection or a complicated disease progression require Authors’ Disclosures of Potential Conflicts of Interest: No authors
validation with a larger group of IE patients. declared any potential conflicts of interest.
The data of our study demonstrated that LBP is Role of Sponsor: The funding organizations played no role in the
strongly correlated with CRP and appears to be a useful design of study, choice of enrolled patients, review and interpretation
tool for the diagnosis of IE and following the response of data, or preparation or approval of manuscript.
of the disease to therapy; however, we have found no Acknowledgments: We thank Sarah Kirkby for her linguistic advice
significant benefit that would legitimize the standard and Doris Hendig for helpful discussions.
References
1. Naber CK, Bauhofer A, Block M, Buerke M, Erbel 2002;89:1400 – 4. with C-reactive protein, interleukin 6 and inter-
R, Graninger W, Herrmann M. S2-Leitlinie zur 10. Rawczynska-Englert I, Hryniewiecki T, Dzierza- feron-␣ for differentiation of bacterial vs. viral
Diagnostik und Therapie der infektiösen Endokar- nowska D. Evaluation of serum cytokine concen- infections. Pediatr Infect Dis J 1999;18:875– 81.
ditis. Z Kardiol 2004;93:1005–21. trations in patients with infective endocarditis. 19. Blairon L, Wittebole X, Laterre PF. Lipopolysac-
2. Durack DT, Lukes AS, Bright DK. New criteria for J Heart Valve Dis 2000;9:705–9. charide-binding protein serum levels in patients
diagnosis of infective endocarditis: utilization of 11. Lamas CC, Eykyn SJ. Suggested modifications to with severe sepsis due to gram-positive and fun-
specific echocardiographic findings. Duke Endo- the Duke criteria for the clinical diagnosis of gal infections. J Infect Dis 2003;187:287–91.
carditis Service. Am J Med 1994;96:200 –9. native valve and prosthetic valve endocarditis: 20. Opal SM, Scannon PJ, Vincent JL, White M, Car-
3. Lisby G, Gutschik E, Durack DT. Molecular meth- analysis of 118 pathologically proven cases. Clin roll SF, Palardy JE, et al. Relationship between
ods for diagnosis of infective endocarditis. Infect Infect Dis 1997;25:713–9. plasma levels of lipopolysaccharide (LPS) and
Dis Clin North Am 2002;16:393– 412. 12. Watkin RW, Harper LV, Vernallis AB, Lang S, LPS-binding protein in patients with severe sepsis
4. Millar B, Moore J, Mallon P, Xu J, Crowe M, Lambert PA, Ranasinghe AM, Elliott TS. Pro- and septic shock. J Infect Dis 1999;180:1584 –9.
McClurg R, et al. Molecular diagnosis of infective inflammatory cytokines IL6, TNF-␣, IL1, procal- 21. Kaden J, Zwerenz P, Lambrecht HG, Dostatni R.
endocarditis—a new Duke’s criterion. Scand J In- citonin, lipopolysaccharide binding protein and
Lipopolysaccharide-binding protein as a new and
fect Dis 2001;33:673– 80. C-reactive protein in infective endocarditis. J In-
reliable infection marker after kidney transplan-
5. Kocazeybek B, Kücükoglu S, Öner YA. Procal- fect 2007;55:220 –5.
tation. Transpl Int 2002;15:163–72.
citonin and C-reactive protein in infective 13. Heiro M, Helenius H, Sundell J, Koskinen P, Eng-
22. Orlikowsky TW, Trug C, Neunhoeffer F, Deper-
endocarditis: correlation with etiology and prog- blom E, Nikoskelainen J, Kotilainen P. Utility of
schmidt M, Eichner M, Poets CF. Lipopolysaccha-
nosis. Chemotherapy 2003;49:76 – 84. serum C-reactive protein in assessing the out-
ride-binding protein in noninfected neonates and
6. Ekdahl C, Broqvist M, Franzen S, Ljunghusen O, come of infective endocarditis. Eur Heart J 2005;
those with suspected early-onset bacterial infec-
Maller R, Sander B. IL-8 and tumor necrosis factor 26:1873– 81.
tion. J Perinatol 2006;26:115–9.
␣ in heart valves from patients with infective 14. Hogevik H, Olaison L, Andersson R, Alestig K.
endocarditis. Scand J Infect Dis 2002;34:759 – 62. C-reactive protein is more sensitive than erythro- 23. Pavcnik-Arnol M, Hojker S, Derganc M. Lipo-
7. Capo C, Zugun F, Stein A, Tardei G, Lepidi H, cyte sedimentation rate for diagnosis of infective polysaccharide-binding protein in critically ill
Raoult D, Mege JL. Upregulation of tumor necro- endocarditis. Infection 1997;25:82–5. neonates and children with suspected infection:
sis factor alpha and interleukin-1 beta in Q fever 15. Olaison L, Hogevik H, Alestig K. Fever, C-reactive comparison with procalcitonin, interleukin-6, and
endocarditis. Infect Immun 1996;64:1638 – 42. protein, and other acute-phase reactants during C-reactive protein. Intensive Care Med 2004;30:
8. Gouriet F, Bothelo-Nevers E, Coulibaly B, Raoult treatment of infective endocarditis. Arch Intern 1454 – 60.
D, Casalta JP. Evaluation of sedimentation rate, Med 1997;157:885–92. 24. Schumann RR, Latz E. Lipopolysaccharide-binding
rheumatoid factor, C-reactive protein, and tumor 16. McCartney AC, Orange GV, Pringle SD, Wills G, protein. Chem Immunol 2000;74:42– 60.
necrosis factor for the diagnosis of infective en- Reece IJ. Serum C reactive protein in infective 25. Tobias PS, Ulevitch RJ. Lipopolysaccharide bind-
docarditis. Clin Vaccine Immunol 2006;13:301. endocarditis. J Clin Pathol 1988;41:44 – 8. ing protein and CD14 in LPS dependent macro-
9. Alter P, Hoeschen J, Ritter M, Maisch B. Useful- 17. Kushner I. C-reactive protein in rheumatology. phage activation. Immunobiology 1993;187:
ness of cytokines interleukin-6 and interleukin-2R Arthritis Rheum 1991;34:1065– 8. 227–32.
concentrations in diagnosing active infective en- 18. Gendrel D, Raymond J, Coste J, Moulin F, Lorrot 26. Schumann RR, Rietschel ET, Loppnow H. The role
docarditis involving native valves. Am J Cardiol M, Guerin S, et al. Comparison of procalcitonin of CD14 and lipopolysaccharide-binding protein