Relationship of Phylogenetic Background, Biofilm Production, and Time to Detection of Growth in Blood Culture Vials with Clinical Variables
http://www.100md.com
微生物临床杂志 2006年第4期
Department of Infectious Diseases
Microbiology Laboratory, Hospital Clínic, IDIBAPS, University of Barcelona, Barcelona, Spain
ABSTRACT
In patients with Escherichia coli bacteremia, data on the relationship of phylogenetic background, biofilm production, and degree of bacteremia with clinical variables and prognosis are scarce. During a 1-year period, all adults with bacteremia due to Escherichia coli diagnosed at a university center were enrolled. Determination of phylogenetic background, biofilm production, and genotyping was performed with all strains, and the time to positivity of blood culture vials was recorded. A total of 185 episodes of diverse-source E. coli bacteremia was analyzed. Strains of phylogroup D were predominant (52%). Phylogroup A isolates were associated with pneumonia and prior antibiotic intake, B1 with an abdominal source of infection, B2 with the absence of urological abnormalities, and D with urological abnormalities and age below 65 years. Resistance to antibiotics and no biofilm production were concentrated in phylogroup A strains. Biofilm production was not associated with any clinical variable. An immunocompromising condition (odds ratio [OR] = 5.01, 95% confidence interval [CI] = 1.4 to 17.9), peritonitis (OR = 17, 95% CI = 3.32 to 87), pneumonia (OR = 9.97, 95% CI = 1.96 to 50.6), and 7 h to bacteremia detection (OR = 4.37, 95% CI = 1.38 to 13.8) were the best predictors of a fatal outcome. Results from this study suggest that the distribution of phylogenetic backgrounds among E. coli strains involved in diverse-source bacteremia may be subject to geographical variation and that, in afflicted individuals, some high-risk sources, the patient's underlying condition, and the degree of bacteremia are more important than microbial factors in determining the outcome. Time to positivity of blood culture vials may be a variable of potential clinical impact.
INTRODUCTION
Escherichia coli continues to be the single most frequently involved gram-negative organism in adult patients with either community-acquired or nosocomial bacteremia (12, 29). Extraintestinal E. coli infections, including bacteremia, are caused predominantly by strains displaying a wide array of putative virulence factors, such as adhesins, iron acquisition systems, host defense-subverting mechanisms, and toxins. The genetic elements encoding these factors have been concentrated predominantly in strains derived from phylogroup B2 and, to a lesser extent, from group D, whereas isolates of groups A and B1 usually contain few virulence determinants and are essentially regarded as commensals (8, 14, 25, 26). Although certain virulence factors have been associated with invasion of specific organs and shedding into the bloodstream (14, 22, 28), there is no proof that they have any impact on prognosis of patients with bacteremia (11). As a matter of fact, mortality associated with bloodstream E. coli infections seems to be influenced mostly by host and treatment factors (10, 21, 27), and patients with local or general compromising conditions are less frequently infected by highly virulent strains (25).
Biofilm formation is commonly associated with infections of medical devices and is considered to be relevant in some spontaneous diseases, such as native valve endocarditis, periodontitis, chronic prostatitis, otitis media, and bronchopulmonary sepsis in patients with cystic fibrosis (7). Although the ability of E. coli to form biofilms is widely recognized, the frequency of this trait in strains causing clinical infection and its relationship to other clinical or bacterial characteristics is still unknown.
Classical studies suggest that the magnitude of bacteremia as determined by quantitative blood cultures may be related to mortality independently of the underlying disease's severity (19). Given the strong relationship between the inoculum size and the time to positivity, as assessed by automatic blood culture-processing devices (2), the latter may constitute a surrogate marker of the degree of bacteremia and provide useful diagnostic and prognostic information (18).
In the present study, we have analyzed the phylogenetic background of the strains involved in a series of unselected and epidemiologically unrelated episodes of E. coli bacteremia, the frequency of biofilm formation in these organisms, and the time to detection of growth as a surrogate marker of the degree of bacteremia in order to gain insight into the relationship of these characteristics with clinical variables and prognosis.
MATERIALS AND METHODS
Patients. The study was conducted at the Hospital Clínic, a 700-bed university hospital in Barcelona, Spain. During one year (2003), all patients with bloodstream infections due to E. coli were prospectively followed up from diagnosis to discharge and the strains involved were assessed for biofilm production, determination of their phylogenetic background, and molecular epidemiological relatedness. This study focuses on episodes of E. coli bacteremia, defined as bloodstream E. coli infections that arose de novo or, in patients with a previous bacteremic E. coli infection, after the apparent clinical cure and source eradication.
Clinical variables. Clinical variables collected from patients with bacteremia included the following: demographics, place of acquisition (hospital or community), primary source of infection, underlying diseases, selected extrinsic risk factors (orotracheal intubation, bladder catheterization, corticosteroids, or other immunosuppressive therapy), selected clinical aspects (fever and shock), antibiotic therapy-related issues (previous administration of antibiotics, ongoing therapy at the time of drawing blood cultures, and empirical and definitive therapies), susceptibility of the infecting strain, and in-hospital mortality. Patients with solid or hematological cancer, human immunodeficiency virus infection, solid organ or bone marrow transplantation, neutropenia, or splenectomy or patients receiving corticosteroids or other immunosuppressive therapy were considered to be immunocompromised. The source of infection was established according to common clinical, analytical, and radiographic criteria and was considered either probable, when cultures from the suspected site were not obtained or negative, or definitive, when an E. coli strain with the same susceptibility pattern as the one isolated from blood was recovered from the apparent source. To consider a venous catheter the source of infection, a positive culture of the catheter tip by the roll plate method and no evidence of other putative origin were required. Antimicrobial therapy given for at least 48 h in the 30 days before bacteremia diagnosis was considered a previous antibiotic treatment. Treatment was considered appropriate when the isolated organism was susceptible in vitro to at least one of the antibiotics given and when the dosage and route of administration were adequate. Detection of growth in a lapse equal to or lower than the 25th percentile of the growth time distribution of the isolates was considered indicative of high-degree bacteremia.
Microbiological procedures. Blood cultures were processed by the BACTEC 9240 system (Becton-Dickinson, Sparks, Md.). When using media without resins, the standard aerobic bottle (BACTEC Standard/10 Aerobic/F) was always paired with the BACTEC Lytic/10 Anaerobic/F bottle. Aerobic and anaerobic resin bottles were BACTEC PLUS Aerobic/F and BACTEC PLUS Anaerobic/F, respectively. The volume of blood cultured was not verified. The exact time from the start of incubation to a positive reading was recorded, considering time to positivity to be that for the bottle in the processed set or sets that became positive first. Antibiotic susceptibility to ampicillin, amoxicillin-clavulanate, cefotaxime, ceftazidime, trimethoprim-sulfamethoxazole, nalidixic acid, ciprofloxacin, and gentamicin was determined by a microdilution method following CLSI (formerly NCCLS) standards (20). The E. coli phylogenetic background was determined by using a three-locus PCR-based method (4). In vitro biofilm formation was analyzed by growing the bacteria into microtiter plates with a minimal glucose medium (M63), stained with a 1% violet crystal solution, and read using a spectrophotometer (6). Molecular epidemiology analysis was performed by pulsed-field gel electrophoresis by using XbaI endonuclease as described elsewhere (9). The subset of phylogroup D strains was serotyped by an agglutination test using specific antiserum from O15, K52, and H1 antigens (Statens Serum Institute, Denmark).
Statistical analysis. Proportions were compared by the 2 test or the Fisher test when appropriate. Quantitative variables were compared by the t test or the Mann-Whitney test. Logistic regression was used to select the best predictors of the phylogenetic background, high-degree bacteremia, and in-hospital mortality. In models predicting mortality, shock was not included as an explanatory variable, due to the fact that it may be a common intermediate condition in the path through which other important prognostic factors lead to death.
RESULTS
One hundred seventy-seven patients had a total of 185 episodes of E. coli bacteremia, of which 96 (52%) had a definitive source. Clinical and microbiological characteristics are shown in Table 1. Almost two-thirds of the episodes occurred in patients older than 65 years, and there was a slight overrepresentation of females. Most individuals suffered from some underlying disease, and more than half of the episodes originated in the urinary tract. Median time to detection of growth by the continuous monitoring blood culture system was 9 h (interquartile interval of 7 to 11 h). Pulsed-field gel electrophoresis confirmed that each isolate corresponded to a unique strain.
Regarding the phylogenetic background of isolates, there was a predominance of strains derived from group D (52%), followed in decreasing order of frequency by those from groups B2 (18%), A (17%), and B1 (12%). The relationship between the phylogroup and clinical or other microbiological variables is shown in Table 2. Strains belonging to group D were the most frequently found among most sources of bacteremia and other clinical characteristics. Only in patients with pneumonia did strains from phylogroup A prevail over group D isolates. Sixteen out of 97 (16%) group D strains belonged to the O15:K52:H1 serotype.
Patients infected with phylogroup A strains more frequently received antibiotics during the last month, had an immunocompromising disease, and had pneumonia as the source of infection. However, only pneumonia (odds ratio [OR] = 4.89, 95% confidence interval [CI] = 1.11 to 21) and prior antibiotic intake (OR = 4.33, 95% CI = 1.88 to 9.95) were independent predictors. Group A strains were also more resistant to any antibiotic except trimethoprim-sulfamethoxazole (data not shown), an association not confounded by recent previous antibiotic exposure. Even for patients who had not received any antibiotic within the last month, rates of resistance to at least one of the drugs tested were significantly higher among phylogroup A strains than among those of other phylogenetic backgrounds (100% versus 74%, respectively; P = 0.01).
Group B1 strains were more frequent in episodes with an abdominal source of infection (peritonitis and biliary tract infections combined) than in those with other origins (OR = 3.01, 95% CI = 1.14 to 9.95). The only independent predictor of group B2 strains was the absence of urological abnormalities (OR = 2.94, 95% CI = 1.04 to 8.33). This association was stronger among patients with a urinary source of infection (OR = 4, 95% CI = 1.02 to 10) than in those without it (OR = 1.75, 95% CI = 0.31 to 17). Group B2 strains were also less resistant to quinolones, an association not confounded by prior quinolone administration. In patients who had not received these drugs within the last month, resistance to nalidixic acid was still lower with B2 strains than with those of other phylogenetic backgrounds (22% versus 42%, respectively; P = 0.04). Group D strains were independently associated with urological abnormalities (OR = 1.96, 95% CI = 1.04 to 3.7) and age below 65 years (OR = 2.48, 95% CI = 1.26 to 4.87). It is of note that the association of group D isolates with urological abnormalities was even stronger for the episodes considered not to have originated in the urinary tract (OR = 3.3, 95% CI = 1.01 to 12; P = 0.03) than for those having a urinary tract source (OR = 1.9, 95% CI = 0.8 to 4.9; P = 0.12). O15:K52:H1 strains did not differ from other group D isolates, except in their higher susceptibility to nalidixic acid (88% versus 54%, respectively; P = 0.01).
Biofilm production was not associated with any demographic or clinical variables and was significantly less common in strains of phylogroup A (5 out of 32, or 16%, for phylogroup A strains versus 74 out of 153, or 48%, for non-phylogroup A strains; P = 0.0007). Strains from patients with bladder catheter-associated urinary tract infection or intravenous catheter-related bacteremia or from men with urinary tract infection without bladder catheters, considered either individually or all together, did not produce biofilm more frequently than strains recovered from other patients (data not shown).
Forty bacteremic episodes were detected by the continuous monitoring blood culture system in 7 h or less of incubation. A univariate analysis of the relationship between time to positivity and clinical or microbiological characteristics is shown in Table 3. Multiple logistic regression identified neutropenia (OR = 4.14, 95% CI = 1.85 to 20.4), a catheter source (OR = 22, 95% CI = 2.22 to 219), liver cirrhosis (OR = 5.06, 95% CI = 1.69 to 15.1), and pneumonia (OR = 6.16, 95% CI = 1.42 to 26.7) as independent predictors of high-degree bacteremia.
Twenty patients (11%) died during hospitalization, and the factors associated with in-hospital mortality are shown in Table 4. Multivariate analysis selected an immunocompromising condition (OR = 5.01, 95% CI = 1.4 to 17.9), peritonitis (OR = 17, 95% CI = 3.32 to 87), pneumonia (OR = 9.97, 95% CI = 1.96 to 50.6), and a time to positivity of 7 h (OR = 4.37, 95% CI = 1.38 to 13.8) as the factors independently associated with a fatal outcome. Appropriate empirical antimicrobial therapy was not associated with a better outcome, and there was no evidence of a significant modifying effect of any clinical or microbiological variable on the association between the appropriateness of initial therapy and mortality.
DISCUSSION
In the present study, we have evaluated the phylogenetic background of the strains involved in a series of unselected and epidemiologically unrelated episodes of diverse-source E. coli bacteremia, the prevalence of biofilm production in those organisms, the time to detection of growth as a surrogate marker of the degree of bacteremia, and the association of these characteristics with several clinical variables and prognosis.
Previous studies have shown that bacteremic E. coli infections in adults from different geographical areas are caused by organisms belonging, in decreasing order of frequency, to phylogroups B2, D, A, and B1. This distribution corresponds with that of the number of putative virulence factors present in those strains (8, 14, 25). In general terms, our data agree with these observations, but some differences are of note. First, strains of group D, instead of group B2, were predominant. Second, a high proportion of strains causing pneumonia derived from phylogroup A, in contrast with previous observations (25). The reasons for these discrepancies are unclear. The relative predominance of different phylogroups may be subject to geographical variations. Fully pathogenic group D strains, such as those belonging to serotype O15:K52:H1, have been described to occur in our geographical area, where they have been endemic for the last decade (5, 23). The prevalence of serotype O15:K52:H1 among bacteremic strains found in this study (8.6%) is at least double the figures previously reported and supports its well-established presence in our region and its increasing involvement in diverse-source bloodstream infections. Regarding the phylogenetic background of strains causing pneumonia, it is conceivable that the relative prevalences of the different groups could vary according to the number of primary versus metastatic lung infections included in different series. Since primary E. coli pneumonia has usually been regarded as caused by commensal organisms belonging to the patient's own flora, there is a possibility that cases due to group A strains may represent true primary pneumonia.
The present data also support the general hypothesis that less virulent strains required some predisposing host conditions to become fully pathogenic and provide some insight on the nature of such precipitating factors. For commensal strains that have become more intrinsically resistant to antibiotics, such as those of group A, recent antibiotic exposure seems to be the most important conditioning event. The relationship between non-B2 phylogenetic background (A or D in particular) and resistance to ciprofloxacin or trimethoprim-sulfamethoxazole or multidrug resistance has previously been recognized (15, 16). The association of B1 strains with an abdominal focus of infection may reflect the overwhelming challenge imposed by either a high inoculum associated with peritonitis or obstruction of the biliary tract. Conversely, presumably virulent B2 strains were associated particularly with bacteremia originating in the urinary tracts of patients without urological abnormalities. This is consistent with previous observations showing the universal predominance of phylogroup B2 among bloodstream or urinary isolates from noncompromised patients with urinary tract infections (1, 13, 17, 32). A plausible explanation for the observed independent association of phylogroup D with younger age and urological abnormalities is, however, not readily apparent. Particularly disturbing is the fact that the association with urological abnormalities was stronger in patients with non-urinary tract infections. Although some misclassification of the actual infection source may have influenced the data, a more plausible reason is that urological abnormalities constitute a surrogate marker for a patient's other conditions, predisposing the patient to bloodstream infections due to relatively virulent strains.
The capacity to produce biofilm was not associated with any clinical characteristic. Even the strains involved in infections that presumably are or may be associated with a biofilm type of growth at the source level, such as venous catheter-related bacteremia, bladder catheter-related urinary tract infection, and urinary tract infection in men without bladder catheters (a population at increased risk for prostatitis), were not more likely biofilm producers than the strains involved in other processes. Although the exact meaning of this observation is uncertain, it suggests that biofilm formation does not foster the ability of E. coli to spread into the bloodstream.
In the present study, we have introduced the time to detection of growth by the continuous monitoring blood culture system as a surrogate marker of the degree of bacteremia and a possible prognostic factor. This measurement has proved to be useful for the diagnosis of venous catheter-related bacteremia (3), and in a recent study, a time to positivity of 14 h was associated with an endovascular source of infection, metastatic infection, and attributable mortality in patients with Staphylococcus aureus bacteremia (18). In fact, three of the factors independently associated with a time to detection equal to or less than 7 h are conditions associated with either severe defects in the mechanisms involved in clearing the blood of microorganisms (neutropenia or liver cirrhosis) or continuous shedding into the bloodstream (catheter-related bacteremia). The association of high-degree bacteremia with pneumonia could reflect the metastatic nature of some of the cases. High-degree bacteremia has been reported to be an important predisposing factor for metastatic shedding of E. coli, particularly in the central nervous system (31). The lack of association of the phylogenetic background with high-degree bacteremia suggests that virulence determinants that promote inflammatory responses, invasion of epithelial cells, and resistance to phagocytosis (such as P fimbriae) may be less relevant in determining the degree of bacteremia (24, 30).
Prognosis of E. coli bacteremia was related to the patient's underlying condition, to some well-known high-risk sources (pneumonia and peritonitis), and to the degree of bacteremia. Neither the phylogenetic background nor the ability to produce biofilm was important in this respect. Present data suggest that the source of infection and host factors, rather than the inferred amount of virulence traits, are the major determinants of prognosis of E. coli bacteremia. This observation agrees with the data from a previous study, in which patients with hemolysin-producing strains in fact had a better prognosis (11). Evidence that high-degree bacteremia as defined in this study was associated with some sources and had prognostic significance may have practical consequences. If these data are confirmed, the microbiology laboratory should consider alerting the physician about any blood culture that becomes positive for a gram-negative bacilli in 7 h. Giving this input on a real-time basis could provide valuable diagnostic information and contribute to the prompt optimization of antimicrobial therapy and other care measures (such as removal of a venous catheter) which may improve prognosis. We do not believe that the lack of association between appropriate initial therapy and prognosis observed in this particular series rules out the potential benefit of optimizing the antimicrobial therapy as soon as possible for patients with bacteremia.
This study has several drawbacks. The most obvious one is that we did not determine the presence of putative virulence factors in the involved strains, and hence we cannot rule out that they actually have prognostic significance, either individually or collectively. Any suggestion in this regard is inferential, based on the well-known association of virulence score with phylogenetic background. On the other hand, only half of the episodes were microbiologically documented at the infection focus and the isolates retrieved from the source were not available for analysis; hence, misclassification of the sources of bacteremia may have biased the results. In addition, we did not verify the volume of blood cultured, assuming that variations occurred randomly. The value of our proposed marker of high-degree bacteremia needs to be confirmed in larger series and with different continuous monitoring blood culture devices.
In summary, the present data suggest that the distribution of phylogenetic backgrounds among strains causing diverse-source E. coli bacteremia may be subject to geographical variation, extend the evidence about the concentration of resistance in group A strains, explore for the first time the clinical significance of biofilm production, and support the concept that prognosis is related mainly to host factors and a putative marker of the degree of bacteremia. Regarding the latter, a time to detection of growth in blood culture vials lower than or equal to the 25th percentile of the time to positivity distribution of all isolates arises as a measurement of possible clinical impact.
ACKNOWLEDGMENTS
We have no conflict of interest regarding the devices, products, or concepts mentioned in the study.
REFERENCES
Bingen-Didois, M., O. Clermont, S. Bonacorsi, M. Terki, N. Brahimi, C. Loukil, D. Barraud, and E. Bingen. 2002. Phylogenetic analysis and prevalence of urosepsis strains of Escherichia coli bearing pathogenicity island-like domains. Infect. Immun. 70:3216-3226.
Blot, F., E. Schmidt, G. Nitenberg, C. Trancrede, B. Leclercq, A. Laplanche, and A. Andremont. 1998. Earlier positivity of central-venous versus peripheral-blood cultures is highly predictive of catheter-related sepsis. J. Clin. Microbiol. 36:105-109.
Blot, F., G. Nitenberg, E. Chachaty, B. Raynard, N. Germann, S. Antoun, A. Laplanche, C. Brun-Buisson, and C. Tancrede. 1999. Diagnosis of catheter-related bacteraemia: a prospective comparison of the time to positivity of hub-blood versus peripheral-blood cultures. Lancet 354:1071-1077.
Clermont, O., S. Bonacorsi, and E. Bingen. 2002. Rapid and simple determination of the Escherichia coli phylogenetic group. Appl. Environ. Microbiol. 66:4555-4558.
Dalmau, D., B. Navarro, J. Mirelis, J. Blanco, J. Garau, and G. Prats. 1996. Escherichia coli bacteremia. Serotype O15:K52:H1 as a urinary pathogen. J. Hosp. Infect. 34:233-234.
Danese, P. N., L. A. Pratt, S. L. Dove, and R. Kolter. 2000. The outer membrane protein, antigen 43, mediates cell-to-cell interactions within Escherichia coli biofilms. Mol. Microbiol. 37:424-432.
Donlan, R. M., and J. W. Costerton. 2002. Biofilms: survival mechanisms of clinically relevant microorganisms. Clin. Microbiol. Rev. 15:167-193.
Duriez, P., O. Clermont, S. Bonacorsi, E. Bingen, A. Chaventre, J. Elion, B. Picard, and E. Denamur. 2001. Commensal Escherichia coli isolates are phylogenetically distributed among geographically distinct human populations. Microbiology 147:1671-1676.
Gautom, R. K. 1997. Rapid pulsed-field gel electrophoresis protocol for typing of Escherichia coli O157:H7 and other gram-negative organisms in 1 day. J. Clin. Microbiol. 35:2977-2980.
Gransden, W. R., S. J. Eykyn, I. Phillips, and B. Rowe. 1990. Bacteremia due to Escherichia coli: a study of 861 episodes. Rev. Infect. Dis. 12:1008-1018.
Hekker, T. A., A. B. Groeneveld, A. M. Simoons-Smit, P. De Man, H. Connell, and D. M. McLaren. 2000. Role of bacterial virulence factors and host factors in the outcome of Escherichia coli bacteremia. Eur. J. Clin. Microbiol. Infect. Dis. 19:312-316.
Javaloyas, M., D. Garcia-Somoza, and F. Gudiol. 2002. Epidemiology and prognosis of bacteremia: a 10-y study in a community hospital. Scand. J. Infect. Dis. 34:436-441.
Johnson, J. R., I. Orskov, F. Orskov, P. Goullet, B. Picard, S. L. Moseley, P. L. Roberts, and W. E. Stamm. 1994. O, K, and H antigens predict virulence factors, carboxylesterase B pattern, antimicrobial resistance, and host compromise among Escherichia coli strains causing urosepsis. J. Infect. Dis. 169:119-126.
Johnson, J. R., M. A. Kuskowski, T. T. O'Bryan, and J. N. Maslow. 2002. Epidemiological correlates of virulence genotype and phylogenetic background among Escherichia coli blood isolates from adults with diverse-source bacteremia. J. Infect. Dis. 185:1439-1447.
Johnson, J. R., M. A. Kuskowski, T. T. O'Bryan, R. Colodner, and R. Raz. 2005. Virulence phenotype and phylogenetic origin in relation to antibiotic resistance profile among Escherichia coli urine sample isolates from Israeli women with acute uncomplicated cystitis. Antimicrob. Agents Chemother. 49:26-31.
Johnson, J. R., M. A. Duskowsli, A. Gajewski, D. F. Sahm, and J. A. Karlowsky. 2005. Virulence characteristics and phylogenetic background of multidrug-resistant and antimicrobial-susceptible isolates of Escherichia coli from across the United States, 2000-2001. J. Infect. Dis. 190:1739-1744.
Johnson, J. R., M. A. Kukowski, A. Gajewski, S. Soto, J. P. Horcajada, M. T. Jimenez de Anta, and J. Vila. 2005. Extended virulence genotypes and phylogenetic background of Escherichia coli isolates from patients with cystitis, pyelonephritis or prostatitis. J. Infect. Dis. 191:46-50.
Khatib, R., K. Riederer, S. Saeed, L. B. Johnson, M. G. Fakih, M. Sharma, M. S. Tabriz, and A. Khosrovaneh. 2005. Time to positivity in Staphylococcus aureus bacteremia: possible correlation with the source and outcome of infection. Clin. Infect. Dis. 41:594-598.
Kreger, B. E., D. E. Craven, P. H. Carling, and W. E. McCabe. 1980. Gram-negative bacteremia. III. Reassessment of etiology, epidemiology and ecology in 612 patients. Am. J. Med. 68:332-343.
National Committee for Clinical Laboratory Standards. 2000. Methods for dilution antimicrobial susceptibility tests for bacteria that grow aerobically. Approved standard M7-A5. National Committee for Clinical Laboratory Standards, Wayne, Pa.
Olesen, B., H. J. Kolmos, F. Orskov, I. Orskov, and A. Gottschau. 1995. Bacteremia due to Escherichia coli in a Danish university hospital, 1986-1990. Scand. J. Infect. Dis. 27:253-257.
Otto, G., T. Sandberg, B. I. Marklund, P. Ulleryd, and C. Svanborg. 1993. Virulence factors and pap genotype in Escherichia coli isolates from women with acute pyelonephritis, with or without bacteremia. Clin. Infect. Dis. 17:448-456.
Prats, G., F. Navarro, B. Mirelis, D. Dalmau, N. Margall, P. Coll, A. Stell, and J. R. Johnson. 2000. Escherichia coli serotype O15:K52:H1 as a uropathogenic clone. J. Clin. Microbiol. 38:201-209.
Rupinder, T., T. Ikeda, R. Malaviya, J. I. MacGregor, J. R. Little, S. J. Hultgren, and S. N. Abraham. 1994. The PapG tip adhesin of P fimbriae protects Escherichia coli from neutrophil bactericidal activity. Infect. Immun. 62:5296-5304.
Russo, T. A., and J. R. Johnson. 2000. Proposal for a new inclusive designation for extraintestinal pathogenic isolates of Escherichia coli: ExPEC. J. Infect. Dis. 181:1753-1754.
Sannes, M. R., M. A. Kuskowski, K. Owens, A. Gajewski, and J. R. Johnson. 2004. Virulence factor profiles and phylogenetic background of Escherichia coli isolates from veterans with bacteremia and uninfected control subjects. J. Infect. Dis. 190:2121-2128.
Vázquez, F., M. C. Mendoza, G. Viejo, and F. J. Mendez. 1992. Survey of Escherichia coli septicemia over a six-year period. Eur. J. Clin. Microbiol. Infect. Dis. 11:110-117.
Wang, M. C., C. C. Tseng, C. Y. Chen, J. J. Wu, and J. J. Huang. 2002. The role of bacterial virulence and host factors in patients with Escherichia coli bacteremia who have acute cholangitis or upper urinary tract infection. Clin. Infect. Dis. 35:1161-1166.
Wisplinghoff, H., T. Bischoff, S. M. Tallent, H. Seifert, R. P. Wenzel, and M. B. Edmond. 2004. Nosocomial bloodstream infections in US hospitals: analysis of 24,179 cases from a prospective nationwide surveillance study. Clin. Infect. Dis. 39:309-317.
Wullt, B., G. Bergsten, H. Connell, P. Rllano, N. Gebratsedlk, L. Hang, and C. Svanborg. 2001. P-fimbriae trigger mucosal responses to Escherichia coli in the human urinary tract. Cell. Microbiol. 3:255-264.
Xie, Y., K. J. Kim, and K. S. Kim. 2004. Current concepts on Escherichia coli K1 translocation of the blood-brain barrier. FEMS Immunol. Med. Microbiol. 42:271-279.
Zhang, K., B. Foxman, and C. Marrs. 2002. Both urinary and rectal Escherichia coli isolates are dominated by strains of phylogenetic group B2. J. Clin. Microbiol. 40:3951-3955.(Jose A. Martínez, Sara So)
Microbiology Laboratory, Hospital Clínic, IDIBAPS, University of Barcelona, Barcelona, Spain
ABSTRACT
In patients with Escherichia coli bacteremia, data on the relationship of phylogenetic background, biofilm production, and degree of bacteremia with clinical variables and prognosis are scarce. During a 1-year period, all adults with bacteremia due to Escherichia coli diagnosed at a university center were enrolled. Determination of phylogenetic background, biofilm production, and genotyping was performed with all strains, and the time to positivity of blood culture vials was recorded. A total of 185 episodes of diverse-source E. coli bacteremia was analyzed. Strains of phylogroup D were predominant (52%). Phylogroup A isolates were associated with pneumonia and prior antibiotic intake, B1 with an abdominal source of infection, B2 with the absence of urological abnormalities, and D with urological abnormalities and age below 65 years. Resistance to antibiotics and no biofilm production were concentrated in phylogroup A strains. Biofilm production was not associated with any clinical variable. An immunocompromising condition (odds ratio [OR] = 5.01, 95% confidence interval [CI] = 1.4 to 17.9), peritonitis (OR = 17, 95% CI = 3.32 to 87), pneumonia (OR = 9.97, 95% CI = 1.96 to 50.6), and 7 h to bacteremia detection (OR = 4.37, 95% CI = 1.38 to 13.8) were the best predictors of a fatal outcome. Results from this study suggest that the distribution of phylogenetic backgrounds among E. coli strains involved in diverse-source bacteremia may be subject to geographical variation and that, in afflicted individuals, some high-risk sources, the patient's underlying condition, and the degree of bacteremia are more important than microbial factors in determining the outcome. Time to positivity of blood culture vials may be a variable of potential clinical impact.
INTRODUCTION
Escherichia coli continues to be the single most frequently involved gram-negative organism in adult patients with either community-acquired or nosocomial bacteremia (12, 29). Extraintestinal E. coli infections, including bacteremia, are caused predominantly by strains displaying a wide array of putative virulence factors, such as adhesins, iron acquisition systems, host defense-subverting mechanisms, and toxins. The genetic elements encoding these factors have been concentrated predominantly in strains derived from phylogroup B2 and, to a lesser extent, from group D, whereas isolates of groups A and B1 usually contain few virulence determinants and are essentially regarded as commensals (8, 14, 25, 26). Although certain virulence factors have been associated with invasion of specific organs and shedding into the bloodstream (14, 22, 28), there is no proof that they have any impact on prognosis of patients with bacteremia (11). As a matter of fact, mortality associated with bloodstream E. coli infections seems to be influenced mostly by host and treatment factors (10, 21, 27), and patients with local or general compromising conditions are less frequently infected by highly virulent strains (25).
Biofilm formation is commonly associated with infections of medical devices and is considered to be relevant in some spontaneous diseases, such as native valve endocarditis, periodontitis, chronic prostatitis, otitis media, and bronchopulmonary sepsis in patients with cystic fibrosis (7). Although the ability of E. coli to form biofilms is widely recognized, the frequency of this trait in strains causing clinical infection and its relationship to other clinical or bacterial characteristics is still unknown.
Classical studies suggest that the magnitude of bacteremia as determined by quantitative blood cultures may be related to mortality independently of the underlying disease's severity (19). Given the strong relationship between the inoculum size and the time to positivity, as assessed by automatic blood culture-processing devices (2), the latter may constitute a surrogate marker of the degree of bacteremia and provide useful diagnostic and prognostic information (18).
In the present study, we have analyzed the phylogenetic background of the strains involved in a series of unselected and epidemiologically unrelated episodes of E. coli bacteremia, the frequency of biofilm formation in these organisms, and the time to detection of growth as a surrogate marker of the degree of bacteremia in order to gain insight into the relationship of these characteristics with clinical variables and prognosis.
MATERIALS AND METHODS
Patients. The study was conducted at the Hospital Clínic, a 700-bed university hospital in Barcelona, Spain. During one year (2003), all patients with bloodstream infections due to E. coli were prospectively followed up from diagnosis to discharge and the strains involved were assessed for biofilm production, determination of their phylogenetic background, and molecular epidemiological relatedness. This study focuses on episodes of E. coli bacteremia, defined as bloodstream E. coli infections that arose de novo or, in patients with a previous bacteremic E. coli infection, after the apparent clinical cure and source eradication.
Clinical variables. Clinical variables collected from patients with bacteremia included the following: demographics, place of acquisition (hospital or community), primary source of infection, underlying diseases, selected extrinsic risk factors (orotracheal intubation, bladder catheterization, corticosteroids, or other immunosuppressive therapy), selected clinical aspects (fever and shock), antibiotic therapy-related issues (previous administration of antibiotics, ongoing therapy at the time of drawing blood cultures, and empirical and definitive therapies), susceptibility of the infecting strain, and in-hospital mortality. Patients with solid or hematological cancer, human immunodeficiency virus infection, solid organ or bone marrow transplantation, neutropenia, or splenectomy or patients receiving corticosteroids or other immunosuppressive therapy were considered to be immunocompromised. The source of infection was established according to common clinical, analytical, and radiographic criteria and was considered either probable, when cultures from the suspected site were not obtained or negative, or definitive, when an E. coli strain with the same susceptibility pattern as the one isolated from blood was recovered from the apparent source. To consider a venous catheter the source of infection, a positive culture of the catheter tip by the roll plate method and no evidence of other putative origin were required. Antimicrobial therapy given for at least 48 h in the 30 days before bacteremia diagnosis was considered a previous antibiotic treatment. Treatment was considered appropriate when the isolated organism was susceptible in vitro to at least one of the antibiotics given and when the dosage and route of administration were adequate. Detection of growth in a lapse equal to or lower than the 25th percentile of the growth time distribution of the isolates was considered indicative of high-degree bacteremia.
Microbiological procedures. Blood cultures were processed by the BACTEC 9240 system (Becton-Dickinson, Sparks, Md.). When using media without resins, the standard aerobic bottle (BACTEC Standard/10 Aerobic/F) was always paired with the BACTEC Lytic/10 Anaerobic/F bottle. Aerobic and anaerobic resin bottles were BACTEC PLUS Aerobic/F and BACTEC PLUS Anaerobic/F, respectively. The volume of blood cultured was not verified. The exact time from the start of incubation to a positive reading was recorded, considering time to positivity to be that for the bottle in the processed set or sets that became positive first. Antibiotic susceptibility to ampicillin, amoxicillin-clavulanate, cefotaxime, ceftazidime, trimethoprim-sulfamethoxazole, nalidixic acid, ciprofloxacin, and gentamicin was determined by a microdilution method following CLSI (formerly NCCLS) standards (20). The E. coli phylogenetic background was determined by using a three-locus PCR-based method (4). In vitro biofilm formation was analyzed by growing the bacteria into microtiter plates with a minimal glucose medium (M63), stained with a 1% violet crystal solution, and read using a spectrophotometer (6). Molecular epidemiology analysis was performed by pulsed-field gel electrophoresis by using XbaI endonuclease as described elsewhere (9). The subset of phylogroup D strains was serotyped by an agglutination test using specific antiserum from O15, K52, and H1 antigens (Statens Serum Institute, Denmark).
Statistical analysis. Proportions were compared by the 2 test or the Fisher test when appropriate. Quantitative variables were compared by the t test or the Mann-Whitney test. Logistic regression was used to select the best predictors of the phylogenetic background, high-degree bacteremia, and in-hospital mortality. In models predicting mortality, shock was not included as an explanatory variable, due to the fact that it may be a common intermediate condition in the path through which other important prognostic factors lead to death.
RESULTS
One hundred seventy-seven patients had a total of 185 episodes of E. coli bacteremia, of which 96 (52%) had a definitive source. Clinical and microbiological characteristics are shown in Table 1. Almost two-thirds of the episodes occurred in patients older than 65 years, and there was a slight overrepresentation of females. Most individuals suffered from some underlying disease, and more than half of the episodes originated in the urinary tract. Median time to detection of growth by the continuous monitoring blood culture system was 9 h (interquartile interval of 7 to 11 h). Pulsed-field gel electrophoresis confirmed that each isolate corresponded to a unique strain.
Regarding the phylogenetic background of isolates, there was a predominance of strains derived from group D (52%), followed in decreasing order of frequency by those from groups B2 (18%), A (17%), and B1 (12%). The relationship between the phylogroup and clinical or other microbiological variables is shown in Table 2. Strains belonging to group D were the most frequently found among most sources of bacteremia and other clinical characteristics. Only in patients with pneumonia did strains from phylogroup A prevail over group D isolates. Sixteen out of 97 (16%) group D strains belonged to the O15:K52:H1 serotype.
Patients infected with phylogroup A strains more frequently received antibiotics during the last month, had an immunocompromising disease, and had pneumonia as the source of infection. However, only pneumonia (odds ratio [OR] = 4.89, 95% confidence interval [CI] = 1.11 to 21) and prior antibiotic intake (OR = 4.33, 95% CI = 1.88 to 9.95) were independent predictors. Group A strains were also more resistant to any antibiotic except trimethoprim-sulfamethoxazole (data not shown), an association not confounded by recent previous antibiotic exposure. Even for patients who had not received any antibiotic within the last month, rates of resistance to at least one of the drugs tested were significantly higher among phylogroup A strains than among those of other phylogenetic backgrounds (100% versus 74%, respectively; P = 0.01).
Group B1 strains were more frequent in episodes with an abdominal source of infection (peritonitis and biliary tract infections combined) than in those with other origins (OR = 3.01, 95% CI = 1.14 to 9.95). The only independent predictor of group B2 strains was the absence of urological abnormalities (OR = 2.94, 95% CI = 1.04 to 8.33). This association was stronger among patients with a urinary source of infection (OR = 4, 95% CI = 1.02 to 10) than in those without it (OR = 1.75, 95% CI = 0.31 to 17). Group B2 strains were also less resistant to quinolones, an association not confounded by prior quinolone administration. In patients who had not received these drugs within the last month, resistance to nalidixic acid was still lower with B2 strains than with those of other phylogenetic backgrounds (22% versus 42%, respectively; P = 0.04). Group D strains were independently associated with urological abnormalities (OR = 1.96, 95% CI = 1.04 to 3.7) and age below 65 years (OR = 2.48, 95% CI = 1.26 to 4.87). It is of note that the association of group D isolates with urological abnormalities was even stronger for the episodes considered not to have originated in the urinary tract (OR = 3.3, 95% CI = 1.01 to 12; P = 0.03) than for those having a urinary tract source (OR = 1.9, 95% CI = 0.8 to 4.9; P = 0.12). O15:K52:H1 strains did not differ from other group D isolates, except in their higher susceptibility to nalidixic acid (88% versus 54%, respectively; P = 0.01).
Biofilm production was not associated with any demographic or clinical variables and was significantly less common in strains of phylogroup A (5 out of 32, or 16%, for phylogroup A strains versus 74 out of 153, or 48%, for non-phylogroup A strains; P = 0.0007). Strains from patients with bladder catheter-associated urinary tract infection or intravenous catheter-related bacteremia or from men with urinary tract infection without bladder catheters, considered either individually or all together, did not produce biofilm more frequently than strains recovered from other patients (data not shown).
Forty bacteremic episodes were detected by the continuous monitoring blood culture system in 7 h or less of incubation. A univariate analysis of the relationship between time to positivity and clinical or microbiological characteristics is shown in Table 3. Multiple logistic regression identified neutropenia (OR = 4.14, 95% CI = 1.85 to 20.4), a catheter source (OR = 22, 95% CI = 2.22 to 219), liver cirrhosis (OR = 5.06, 95% CI = 1.69 to 15.1), and pneumonia (OR = 6.16, 95% CI = 1.42 to 26.7) as independent predictors of high-degree bacteremia.
Twenty patients (11%) died during hospitalization, and the factors associated with in-hospital mortality are shown in Table 4. Multivariate analysis selected an immunocompromising condition (OR = 5.01, 95% CI = 1.4 to 17.9), peritonitis (OR = 17, 95% CI = 3.32 to 87), pneumonia (OR = 9.97, 95% CI = 1.96 to 50.6), and a time to positivity of 7 h (OR = 4.37, 95% CI = 1.38 to 13.8) as the factors independently associated with a fatal outcome. Appropriate empirical antimicrobial therapy was not associated with a better outcome, and there was no evidence of a significant modifying effect of any clinical or microbiological variable on the association between the appropriateness of initial therapy and mortality.
DISCUSSION
In the present study, we have evaluated the phylogenetic background of the strains involved in a series of unselected and epidemiologically unrelated episodes of diverse-source E. coli bacteremia, the prevalence of biofilm production in those organisms, the time to detection of growth as a surrogate marker of the degree of bacteremia, and the association of these characteristics with several clinical variables and prognosis.
Previous studies have shown that bacteremic E. coli infections in adults from different geographical areas are caused by organisms belonging, in decreasing order of frequency, to phylogroups B2, D, A, and B1. This distribution corresponds with that of the number of putative virulence factors present in those strains (8, 14, 25). In general terms, our data agree with these observations, but some differences are of note. First, strains of group D, instead of group B2, were predominant. Second, a high proportion of strains causing pneumonia derived from phylogroup A, in contrast with previous observations (25). The reasons for these discrepancies are unclear. The relative predominance of different phylogroups may be subject to geographical variations. Fully pathogenic group D strains, such as those belonging to serotype O15:K52:H1, have been described to occur in our geographical area, where they have been endemic for the last decade (5, 23). The prevalence of serotype O15:K52:H1 among bacteremic strains found in this study (8.6%) is at least double the figures previously reported and supports its well-established presence in our region and its increasing involvement in diverse-source bloodstream infections. Regarding the phylogenetic background of strains causing pneumonia, it is conceivable that the relative prevalences of the different groups could vary according to the number of primary versus metastatic lung infections included in different series. Since primary E. coli pneumonia has usually been regarded as caused by commensal organisms belonging to the patient's own flora, there is a possibility that cases due to group A strains may represent true primary pneumonia.
The present data also support the general hypothesis that less virulent strains required some predisposing host conditions to become fully pathogenic and provide some insight on the nature of such precipitating factors. For commensal strains that have become more intrinsically resistant to antibiotics, such as those of group A, recent antibiotic exposure seems to be the most important conditioning event. The relationship between non-B2 phylogenetic background (A or D in particular) and resistance to ciprofloxacin or trimethoprim-sulfamethoxazole or multidrug resistance has previously been recognized (15, 16). The association of B1 strains with an abdominal focus of infection may reflect the overwhelming challenge imposed by either a high inoculum associated with peritonitis or obstruction of the biliary tract. Conversely, presumably virulent B2 strains were associated particularly with bacteremia originating in the urinary tracts of patients without urological abnormalities. This is consistent with previous observations showing the universal predominance of phylogroup B2 among bloodstream or urinary isolates from noncompromised patients with urinary tract infections (1, 13, 17, 32). A plausible explanation for the observed independent association of phylogroup D with younger age and urological abnormalities is, however, not readily apparent. Particularly disturbing is the fact that the association with urological abnormalities was stronger in patients with non-urinary tract infections. Although some misclassification of the actual infection source may have influenced the data, a more plausible reason is that urological abnormalities constitute a surrogate marker for a patient's other conditions, predisposing the patient to bloodstream infections due to relatively virulent strains.
The capacity to produce biofilm was not associated with any clinical characteristic. Even the strains involved in infections that presumably are or may be associated with a biofilm type of growth at the source level, such as venous catheter-related bacteremia, bladder catheter-related urinary tract infection, and urinary tract infection in men without bladder catheters (a population at increased risk for prostatitis), were not more likely biofilm producers than the strains involved in other processes. Although the exact meaning of this observation is uncertain, it suggests that biofilm formation does not foster the ability of E. coli to spread into the bloodstream.
In the present study, we have introduced the time to detection of growth by the continuous monitoring blood culture system as a surrogate marker of the degree of bacteremia and a possible prognostic factor. This measurement has proved to be useful for the diagnosis of venous catheter-related bacteremia (3), and in a recent study, a time to positivity of 14 h was associated with an endovascular source of infection, metastatic infection, and attributable mortality in patients with Staphylococcus aureus bacteremia (18). In fact, three of the factors independently associated with a time to detection equal to or less than 7 h are conditions associated with either severe defects in the mechanisms involved in clearing the blood of microorganisms (neutropenia or liver cirrhosis) or continuous shedding into the bloodstream (catheter-related bacteremia). The association of high-degree bacteremia with pneumonia could reflect the metastatic nature of some of the cases. High-degree bacteremia has been reported to be an important predisposing factor for metastatic shedding of E. coli, particularly in the central nervous system (31). The lack of association of the phylogenetic background with high-degree bacteremia suggests that virulence determinants that promote inflammatory responses, invasion of epithelial cells, and resistance to phagocytosis (such as P fimbriae) may be less relevant in determining the degree of bacteremia (24, 30).
Prognosis of E. coli bacteremia was related to the patient's underlying condition, to some well-known high-risk sources (pneumonia and peritonitis), and to the degree of bacteremia. Neither the phylogenetic background nor the ability to produce biofilm was important in this respect. Present data suggest that the source of infection and host factors, rather than the inferred amount of virulence traits, are the major determinants of prognosis of E. coli bacteremia. This observation agrees with the data from a previous study, in which patients with hemolysin-producing strains in fact had a better prognosis (11). Evidence that high-degree bacteremia as defined in this study was associated with some sources and had prognostic significance may have practical consequences. If these data are confirmed, the microbiology laboratory should consider alerting the physician about any blood culture that becomes positive for a gram-negative bacilli in 7 h. Giving this input on a real-time basis could provide valuable diagnostic information and contribute to the prompt optimization of antimicrobial therapy and other care measures (such as removal of a venous catheter) which may improve prognosis. We do not believe that the lack of association between appropriate initial therapy and prognosis observed in this particular series rules out the potential benefit of optimizing the antimicrobial therapy as soon as possible for patients with bacteremia.
This study has several drawbacks. The most obvious one is that we did not determine the presence of putative virulence factors in the involved strains, and hence we cannot rule out that they actually have prognostic significance, either individually or collectively. Any suggestion in this regard is inferential, based on the well-known association of virulence score with phylogenetic background. On the other hand, only half of the episodes were microbiologically documented at the infection focus and the isolates retrieved from the source were not available for analysis; hence, misclassification of the sources of bacteremia may have biased the results. In addition, we did not verify the volume of blood cultured, assuming that variations occurred randomly. The value of our proposed marker of high-degree bacteremia needs to be confirmed in larger series and with different continuous monitoring blood culture devices.
In summary, the present data suggest that the distribution of phylogenetic backgrounds among strains causing diverse-source E. coli bacteremia may be subject to geographical variation, extend the evidence about the concentration of resistance in group A strains, explore for the first time the clinical significance of biofilm production, and support the concept that prognosis is related mainly to host factors and a putative marker of the degree of bacteremia. Regarding the latter, a time to detection of growth in blood culture vials lower than or equal to the 25th percentile of the time to positivity distribution of all isolates arises as a measurement of possible clinical impact.
ACKNOWLEDGMENTS
We have no conflict of interest regarding the devices, products, or concepts mentioned in the study.
REFERENCES
Bingen-Didois, M., O. Clermont, S. Bonacorsi, M. Terki, N. Brahimi, C. Loukil, D. Barraud, and E. Bingen. 2002. Phylogenetic analysis and prevalence of urosepsis strains of Escherichia coli bearing pathogenicity island-like domains. Infect. Immun. 70:3216-3226.
Blot, F., E. Schmidt, G. Nitenberg, C. Trancrede, B. Leclercq, A. Laplanche, and A. Andremont. 1998. Earlier positivity of central-venous versus peripheral-blood cultures is highly predictive of catheter-related sepsis. J. Clin. Microbiol. 36:105-109.
Blot, F., G. Nitenberg, E. Chachaty, B. Raynard, N. Germann, S. Antoun, A. Laplanche, C. Brun-Buisson, and C. Tancrede. 1999. Diagnosis of catheter-related bacteraemia: a prospective comparison of the time to positivity of hub-blood versus peripheral-blood cultures. Lancet 354:1071-1077.
Clermont, O., S. Bonacorsi, and E. Bingen. 2002. Rapid and simple determination of the Escherichia coli phylogenetic group. Appl. Environ. Microbiol. 66:4555-4558.
Dalmau, D., B. Navarro, J. Mirelis, J. Blanco, J. Garau, and G. Prats. 1996. Escherichia coli bacteremia. Serotype O15:K52:H1 as a urinary pathogen. J. Hosp. Infect. 34:233-234.
Danese, P. N., L. A. Pratt, S. L. Dove, and R. Kolter. 2000. The outer membrane protein, antigen 43, mediates cell-to-cell interactions within Escherichia coli biofilms. Mol. Microbiol. 37:424-432.
Donlan, R. M., and J. W. Costerton. 2002. Biofilms: survival mechanisms of clinically relevant microorganisms. Clin. Microbiol. Rev. 15:167-193.
Duriez, P., O. Clermont, S. Bonacorsi, E. Bingen, A. Chaventre, J. Elion, B. Picard, and E. Denamur. 2001. Commensal Escherichia coli isolates are phylogenetically distributed among geographically distinct human populations. Microbiology 147:1671-1676.
Gautom, R. K. 1997. Rapid pulsed-field gel electrophoresis protocol for typing of Escherichia coli O157:H7 and other gram-negative organisms in 1 day. J. Clin. Microbiol. 35:2977-2980.
Gransden, W. R., S. J. Eykyn, I. Phillips, and B. Rowe. 1990. Bacteremia due to Escherichia coli: a study of 861 episodes. Rev. Infect. Dis. 12:1008-1018.
Hekker, T. A., A. B. Groeneveld, A. M. Simoons-Smit, P. De Man, H. Connell, and D. M. McLaren. 2000. Role of bacterial virulence factors and host factors in the outcome of Escherichia coli bacteremia. Eur. J. Clin. Microbiol. Infect. Dis. 19:312-316.
Javaloyas, M., D. Garcia-Somoza, and F. Gudiol. 2002. Epidemiology and prognosis of bacteremia: a 10-y study in a community hospital. Scand. J. Infect. Dis. 34:436-441.
Johnson, J. R., I. Orskov, F. Orskov, P. Goullet, B. Picard, S. L. Moseley, P. L. Roberts, and W. E. Stamm. 1994. O, K, and H antigens predict virulence factors, carboxylesterase B pattern, antimicrobial resistance, and host compromise among Escherichia coli strains causing urosepsis. J. Infect. Dis. 169:119-126.
Johnson, J. R., M. A. Kuskowski, T. T. O'Bryan, and J. N. Maslow. 2002. Epidemiological correlates of virulence genotype and phylogenetic background among Escherichia coli blood isolates from adults with diverse-source bacteremia. J. Infect. Dis. 185:1439-1447.
Johnson, J. R., M. A. Kuskowski, T. T. O'Bryan, R. Colodner, and R. Raz. 2005. Virulence phenotype and phylogenetic origin in relation to antibiotic resistance profile among Escherichia coli urine sample isolates from Israeli women with acute uncomplicated cystitis. Antimicrob. Agents Chemother. 49:26-31.
Johnson, J. R., M. A. Duskowsli, A. Gajewski, D. F. Sahm, and J. A. Karlowsky. 2005. Virulence characteristics and phylogenetic background of multidrug-resistant and antimicrobial-susceptible isolates of Escherichia coli from across the United States, 2000-2001. J. Infect. Dis. 190:1739-1744.
Johnson, J. R., M. A. Kukowski, A. Gajewski, S. Soto, J. P. Horcajada, M. T. Jimenez de Anta, and J. Vila. 2005. Extended virulence genotypes and phylogenetic background of Escherichia coli isolates from patients with cystitis, pyelonephritis or prostatitis. J. Infect. Dis. 191:46-50.
Khatib, R., K. Riederer, S. Saeed, L. B. Johnson, M. G. Fakih, M. Sharma, M. S. Tabriz, and A. Khosrovaneh. 2005. Time to positivity in Staphylococcus aureus bacteremia: possible correlation with the source and outcome of infection. Clin. Infect. Dis. 41:594-598.
Kreger, B. E., D. E. Craven, P. H. Carling, and W. E. McCabe. 1980. Gram-negative bacteremia. III. Reassessment of etiology, epidemiology and ecology in 612 patients. Am. J. Med. 68:332-343.
National Committee for Clinical Laboratory Standards. 2000. Methods for dilution antimicrobial susceptibility tests for bacteria that grow aerobically. Approved standard M7-A5. National Committee for Clinical Laboratory Standards, Wayne, Pa.
Olesen, B., H. J. Kolmos, F. Orskov, I. Orskov, and A. Gottschau. 1995. Bacteremia due to Escherichia coli in a Danish university hospital, 1986-1990. Scand. J. Infect. Dis. 27:253-257.
Otto, G., T. Sandberg, B. I. Marklund, P. Ulleryd, and C. Svanborg. 1993. Virulence factors and pap genotype in Escherichia coli isolates from women with acute pyelonephritis, with or without bacteremia. Clin. Infect. Dis. 17:448-456.
Prats, G., F. Navarro, B. Mirelis, D. Dalmau, N. Margall, P. Coll, A. Stell, and J. R. Johnson. 2000. Escherichia coli serotype O15:K52:H1 as a uropathogenic clone. J. Clin. Microbiol. 38:201-209.
Rupinder, T., T. Ikeda, R. Malaviya, J. I. MacGregor, J. R. Little, S. J. Hultgren, and S. N. Abraham. 1994. The PapG tip adhesin of P fimbriae protects Escherichia coli from neutrophil bactericidal activity. Infect. Immun. 62:5296-5304.
Russo, T. A., and J. R. Johnson. 2000. Proposal for a new inclusive designation for extraintestinal pathogenic isolates of Escherichia coli: ExPEC. J. Infect. Dis. 181:1753-1754.
Sannes, M. R., M. A. Kuskowski, K. Owens, A. Gajewski, and J. R. Johnson. 2004. Virulence factor profiles and phylogenetic background of Escherichia coli isolates from veterans with bacteremia and uninfected control subjects. J. Infect. Dis. 190:2121-2128.
Vázquez, F., M. C. Mendoza, G. Viejo, and F. J. Mendez. 1992. Survey of Escherichia coli septicemia over a six-year period. Eur. J. Clin. Microbiol. Infect. Dis. 11:110-117.
Wang, M. C., C. C. Tseng, C. Y. Chen, J. J. Wu, and J. J. Huang. 2002. The role of bacterial virulence and host factors in patients with Escherichia coli bacteremia who have acute cholangitis or upper urinary tract infection. Clin. Infect. Dis. 35:1161-1166.
Wisplinghoff, H., T. Bischoff, S. M. Tallent, H. Seifert, R. P. Wenzel, and M. B. Edmond. 2004. Nosocomial bloodstream infections in US hospitals: analysis of 24,179 cases from a prospective nationwide surveillance study. Clin. Infect. Dis. 39:309-317.
Wullt, B., G. Bergsten, H. Connell, P. Rllano, N. Gebratsedlk, L. Hang, and C. Svanborg. 2001. P-fimbriae trigger mucosal responses to Escherichia coli in the human urinary tract. Cell. Microbiol. 3:255-264.
Xie, Y., K. J. Kim, and K. S. Kim. 2004. Current concepts on Escherichia coli K1 translocation of the blood-brain barrier. FEMS Immunol. Med. Microbiol. 42:271-279.
Zhang, K., B. Foxman, and C. Marrs. 2002. Both urinary and rectal Escherichia coli isolates are dominated by strains of phylogenetic group B2. J. Clin. Microbiol. 40:3951-3955.(Jose A. Martínez, Sara So)