Phylogenetic and Pathotypic Comparison of Concurrent Urine and Rectal Escherichia coli Isolates from Men with Febrile Urinary Tract Infectio
http://www.100md.com
微生物临床杂志 2005年第8期
Mucosal and Vaccine Research Center
Geriatric Research, Education, and Clinical Center, Minneapolis VA Medical Center
Departments of Medicine
Psychiatry, University of Minnesota, Minneapolis, Minnesota
International Escherichia coli and Klebsiella Center, Statens Seruminstitut, Copenhagen, Denmark
Department of Infectious Diseases, University of Gteborg, Gteborg, Sweden
ABSTRACT
Among men with febrile urinary tract infection (FUTI), whether the host's fecal flora is the source for the urine strain ("fecal-urethral" hypothesis), and whether pathogenesis is driven by prevalence versus special pathogenicity, are unknown. Accordingly, pretherapy urine isolates from 65 men with FUTI were compared with concurrent rectal isolates from the same hosts according to serotype, genomic profile, phylogenetic group, and virulence genotype. The host's multiple rectal colonies included only the urine clone in 25% of subjects, the urine clone plus additional clones in 22%, and only nonurine clones in 54%. Compared with the 67 unique rectal clones, the 65 urine isolates were significantly enriched for phylogenetic group B2, virulence-associated serotypes, and specific virulence genes and contained more virulence genes (median, 10 versus 6: P < 0.001). In multivariable models, phylogenetic group B2, hlyD (hemolysin), cnf1 (cytotoxic necrotizing factor), iroN (siderophore receptor), ompT (outer membrane protease), and malX (pathogenicity island marker) most strongly predicted urine source. These findings challenge the fecal-urethral and prevalence hypotheses for FUTI pathogenesis and instead strongly support the possibility of alternate infection routes in some men and the special pathogenicity hypothesis. They also identify specific bacterial traits as potential targets for anti-FUTI interventions.
INTRODUCTION
Escherichia coli is a major extraintestinal pathogen, causing tremendous morbidity, mortality, and increased health care costs (33). Urinary tract infection (UTI), the principal extraintestinal syndrome caused by E. coli, although most common among women, is also a significant problem among men, who can experience cystitis, pyelonephritis, acute and chronic prostatitis, and febrile UTI (FUTI) (23, 24, 41). Better understanding of the pathogenesis of UTI in men is needed to guide the development of effective preventive measures.
It is unknown whether in men with UTI the causative E. coli strains usually derive immediately from the host's own intestinal flora, as is true in females, according to the "fecal-urethral" hypothesis (9, 44). For example, vaginal or anal intercourse may introduce microorganisms directly into a man's urethra from his sex partner's vaginal or rectal flora, without involving his intestinal tract (2, 3, 8, 43). Likewise, it is not known whether the particular E. coli strains that cause UTI in men do so merely because of their high prevalence within the host's intestinal flora, consistent with the "prevalence" hypothesis for UTI pathogenesis (38), or instead because they possess an enhanced ability to cause extraintestinal disease compared with other E. coli strains, consistent with the "special pathogenicity" hypothesis that is thought to underlie UTI pathogenesis in most females (27, 30). Although the virulence factors (VF) and surface antigens of E. coli isolates from men with febrile UTI and prostatitis exhibit many similarities to those of strains causing pyelonephritis in women and girls, consistent with the "special pathogenicity" hypothesis (1, 25, 32, 37, 39), few data are available comparing male UTI isolates with fecal isolates from the same hosts (36).
Accordingly, we compared a collection of E. coli urine isolates from a cohort of 65 men with FUTI (a subset of the 70 isolates described in a recent report [17]) with concurrent pretherapy fecal E. coli isolates from the same men to assess both the "fecal-urethral" pathway and the "prevalence" versus "special pathogenicity" hypotheses in the context of FUTI in men. Specifically, we used genomic profiling to resolve distinct clones among the subjects' multiple fecal isolates and then compared urine and fecal clones with respect to surface antigens, phylogenetic background, and VF profiles.
MATERIALS AND METHODS
Subjects and isolates. The 65 men with FUTI due to E. coli were prospectively enrolled at Sahlgrenska University Hospital, Gteborg, Sweden, between March 1993 and December 1996, as reported in greater detail elsewhere (17). Briefly, to be included, patients had to have a fever of 38.0°C and at least one symptom or sign referable to the urinary tract (frequency, dysuria, flank pain, or costovertebral angle tenderness), as well as 104 CFU/ml of E. coli in midstream urine. No patient presented with classical acute prostatitis, although eight had slight prostatic tenderness. None developed symptoms while in hospital. All patients were hospitalized for at least 1 day. Patients with a chronic indwelling urinary catheter were excluded. No patient had known human immunodeficiency virus infection. The University of Gteborg Research Ethics Committee approved the study protocol. All subjects provided informed consent to participate.
Signs and symptoms, a history of previous UTI and genitourinary disorders, concomitant diseases, and a clinical evaluation of the patient's general condition were recorded. Urine, blood samples, and a rectal swab for culture were obtained before antimicrobial treatment. The serum prostate-specific antigen (PSA) level was measured (normal, 4.0 ng/ml) (41). Uroflowmetry, digital rectal examination, and measurement of postvoid residual urine by abdominal ultrasound were done at entry and after 3 months. Cystourethroscopy and imaging studies of the upper urinary tract were performed about 3 months after enrollment, unless carried out within the last 5 years (40).
Urine and rectal cultures. Urine was cultured semiquantitatively on blood agar and cysteine-lactose-electrolyte-deficient agar plates using a calibrated loop technique. A single colony was saved for further analysis. Rectal swabs were cultured on modified Drigalski agar plates (22). From each plate, the last three colonies at the end of the streak area, plus any other morphologically distinct colonies, were saved for further analysis. This method provides 97% probability of capturing the quantitatively predominant clone in the sample (22). The 65 subjects described here represent all of the previously described 70 men with E. coli FUTI (17) from whom rectal E. coli isolates were available for analysis. (For the remaining five subjects, the initial rectal isolates either were not saved or were not retrievable from storage.)
Selection and clonal analysis of urine and rectal isolates. Each host's urine isolate and multiple (3) fecal isolates underwent O:K:H (somatic, capsule, and flagellar) antigen detection, according to standard methods (28). For the fecal isolates from 10 subjects, a truncated serotyping approach was used in which K and H antigens were determined only if there was an O-antigen match with the host's urine isolate.
To resolve discrete clones, all isolates (urine and rectal) from a given host were compared according to genomic profiles as generated using random amplified polymorphic DNA (RAPD) analysis. Random oligomer 1247 was used, supplemented in some instances with oligomer 1290, with visual inspection of profiles in ethidium bromide-stained agarose gels used to define unique clones (4). Serologically similar isolates that exhibited indistinguishable RAPD profiles were defined as representing the same clone, whereas serologically distinct isolates that exhibited distinct RAPD profiles were defined as representing different clones. Ambiguous results, and any discrepancies between the serological data and RAPD analysis, were further assessed by pulsed-field gel electrophoresis (5), with a three-band difference used as the threshold criterion for defining clonality (35).
Phylotyping and virulence genotyping. All urine isolates and at least one representative of each unique clone among each host's rectal isolates underwent PCR-based testing for major E. coli phylogenetic group (A, B1, B2, and D) (6), extended virulence genotype (for 34 markers) (19), and, if positive for any pap (P fimbriae) element, 12 alleles of papA (P fimbriae structural subunit) (20). All assays were done at least in duplicate using independently prepared lysates of the test strains and appropriate positive and negative controls. The VF score was the sum of virulence markers detected in an isolate, adjusted for multiple detection of the pap, sfa/foc (S/F1C fimbriae), and kps II (group 2 capsule) operons.
Statistical methods. Hosts and the corresponding urine isolates were stratified for analysis into three groups, according to the degree of commonality among the host's urine and rectal clones. Group 1 comprised hosts for whom all rectal isolates represented the urine clone, group 2 those whose rectal isolates included the urine clone but also at least one nonurine clone, and group 3 those with only nonurine rectal clones. Host characteristics and acute manifestations of FUTI that were analyzed included bacteremia, flank pain, acute PSA level (absolute and categorically as >4 ng/ml), host age (absolute and categorically as 45, 50, and >65), previous UTI, previous FUTI, increased postvoid residual volume, recent urinary tract instrumentation, renal cortical scarring, diabetes mellitus, and a combination of (age >65, elevated PSA level, and 1 possibly predisposing condition), as previously described in greater detail (17). For analytical purposes, 10 UTI-associated O serogroups, i.e., O1, O2, O4, O6, O7, O8, O16, O18, O25, and O75, were defined as the O-UTI groups (16, 29), six UTI-associated K antigens, i.e., K1, K2, K3, K5, K12, and K13, were defined as the K-UTI types (16), and 12 pyelonephritis-associated O:K:H serotypes, i.e., O1:K1:H7, O2:K1:H4, O4:K5:H5, O4:K12:H1, O4:K12:H5, O6:K2:H1, O6:K5:H1, O7:K1:H1, O16:K1:H6, O18:K1:H7, O18:K5:H–, and O18:K5:H7, were defined as the O:K:H-Pyelo serotypes (16, 29). Isolates conforming to one of the O:K:H-Pyelo serotypes except for an absent K or H antigen were analyzed as belonging to that serotype (16).
Comparisons of proportions were tested using Fisher's exact test (two-tailed). Comparisons of continuous variables were tested using the Mann-Whitney U test (exact). Odds ratios (ORs), with 95% confidence intervals (CIs), for individual or multiple variables as predictors of urine versus rectal origin were calculated using, respectively, univariate or stepwise multiple logistic regression analysis. Because of multiple comparisons, the significance criterion was a P value of 0.01.
RESULTS
Clonal analysis of urine and rectal isolates. According to O:K:H serotyping, RAPD analysis, and selective pulsed-field gel electrophoresis analysis of concurrent pretherapy urine and rectal isolates from 65 men with FUTI, for 16 (25%) of the subjects (group 1), the host's urine clone accounted for all of the multiple rectal isolates examined. In contrast, for 49 (75%) of the subjects, the rectal isolates included one or more distinct (nonurine) clones, either together with the host's urine clone (group 2; n = 14 [22%]) or in the absence of the urine clone (group 3; n = 35 [54%]). Thus, the urine clone clearly predominated within the host's rectal flora in 25% of subjects (group 1), clearly did not predominate there in 54% of subjects (group 3), and may or may not have predominated in 22% of subjects (group 2). Overall, the host's urine clone was identified within the host's rectal flora for 30 (46%) of the 65 subjects (groups 1 and 2). The 49 subjects (76%) from groups 2 and 3 each had from 1 to 3 unique (nonurine) rectal clones identified, giving a total of 67 unique rectal clones for comparison with the 65 urine clones.
Urine strain characteristics versus host colonization status. If high prevalence within the rectal flora could allow a less-virulent clone to cause FUTI, then the host's rectal colonization status with respect to the urine clone (versus other E. coli clones) should influence the characteristics of the urine clone. Therefore, comparisons were made among the 65 FUTI urine isolates according to 53 bacterial characteristics (serological categories, phylogenetic groups, and virulence genes) in relation to rectal colonization status. In comparisons of group 1 versus groups 2 and 3 combined, group 3 versus groups 1 and 2 combined, and group 1 versus group 3, the resulting 159 comparisons yielded no statistically significant differences and only two borderline significant differences (for both, P = 0.03), which could have arisen by chance alone. This provided evidence against prevalence within the rectal flora as a significant determinant of bacterial characteristics in FUTI and supported combining the urine isolates as an "all urine" group.
Host characteristics and acute manifestations according to rectal colonization status. Analogous to the minimal differences noted according to bacterial characteristics, minimal differences also were detected among the 3 host subgroups when they were compared for host characteristics and acute manifestations of FUTI (16 variables total), using the same groupings as described above. Only 1 of the 48 resulting comparisons yielded a statistically significant association, and 2 yielded borderline significant associations, all involving age. That is, group 3 hosts (who had only nonurine rectal clones) were somewhat less likely to be >65 years old (8/35; 23%) than were either group 1 hosts (9/16; 56%; P = 0.03) or all other hosts combined (17/30; 57%; P = 0.01) and were somewhat younger in absolute terms (median age, 55 years [18 to 86] than other hosts (median age, 69 years [32 to 85] (P = 0.025). Since these associations were not particularly strong and, in view of the multiple comparisons, could have arisen by chance alone, the three subgroups of urine isolates were combined as an "all urine" group for enhanced statistical power in comparisons versus the unique rectal isolates.
Urine isolates versus unique rectal strains. Compared with the 67 unique rectal strains, the 65 urine isolates were strikingly enriched for phylogenetic group B2 and most of the studied VFs and virulence-associated serological categories (Table 1). Accordingly, aggregate VF scores were substantially higher among the urine isolates than among the rectal isolates (median score, 10 versus 6: P < 0.001) (Fig. 1). Likewise, among the 12 papA alleles the urine isolates exhibited a significantly greater prevalence of the F10 variant (25% versus 6%: P = 0.002) and similar borderline trends for the F16 and F48 variants (P = 0.03 for both). Similar results, albeit with slightly higher P values due to the smaller sample size, were obtained when, for optimal matching, the analysis was limited to those urine isolates from the 49 hosts (groups 2 and 3) who contributed 1 unique rectal clone (not shown).
Because of the multiplicity of traits associated with urine origin by univariate analysis, stepwise multiple logistic regression analysis was used to identify independent predictors of urine source. A series of models was constructed using different groups of molecularly defined bacterial traits as candidate predictor variables (Table 2). The variables that in one or more of these models significantly predicted urine source included phylogenetic group B2 (the strongest predictor overall), hlyD, ompT, malX, and cnf1 (Table 2). In contrast, in two models the F12 papA allele was a significant negative predictor of urine source (Table 2).
These analyses were repeated with the three serological categories (i.e., O-UTI, K-UTI, and O:K:H-pyelo) as added candidate predictor variables, which reduced the sample size from 132 to 117 because of incomplete serological data for 15 fecal isolates. The results were largely unchanged overall, and no serological category emerged as a significant predictor in any model (not shown). Differences compared with Table 2 included, in model 1, a step 4 in which iroN emerged as another significant positive predictor of urine source (OR, 7.6 [CI, 2.0 to 29.4]; P = 0.003) and a step 5 in which ibeA also emerged as a significant negative predictor (OR, 0.1 [CI, 0.02 to 0.6]; P = 0.009). Additionally, in models 2 and 3, iroN (OR, 13.0 [CI, 3.6 to 46.1] in model 2, and OR, 6.5 [CI 2.5 to 17.4] in model 3; P < 0.001 for both) replaced ompT as a positive predictor, whereas model 2 stopped after step 2 (not shown).
To clarify the basis for the strong predictive power of phylogenetic group B2, the phylogenetic distribution of VFs was assessed among the combined urine and rectal isolates (n = 132). Nearly all of the VFs that were significantly associated with urine source (Table 1) were also highly associated with group B2 (not shown). Accordingly, VF scores were much higher for group B2 than for the other phylogenetic groups combined (median, 10 [range, 2 to 15], versus. 3.5 [range, 0 to 12]; P < 0.001) (Fig. 1). This relationship held independently of urine versus rectal source. That is, urine and rectal source group B2 isolates exhibited similarly high median VF scores (10 in both groups), although the range extended lower among the rectal isolates (urine, [8 to 14]; rectal, [2 to 15]; P = 0.04). Likewise, urine and rectal source non-B2 isolates exhibited similarly low VF scores (median, 3.0 [1 to 8] versus 3.5 [0 to 12]; P > 0.10).
DISCUSSION
In this molecular epidemiological and phylogenetic comparison of urine and concurrent rectal E. coli isolates from men with FUTI, we found that the host's urine clone predominated within the host's gut flora at the time of presentation in only a minority of subjects. Moreover, those urine clones that did predominate rectally were essentially indistinguishable from those that did not, as were the corresponding hosts, except for possible minor age differences. In contrast, urine clones collectively differed dramatically from rectal clones with respect to phylogenetic background, virulence gene content, and serotypes, in patterns suggesting much greater virulence potential for the urine isolates. These findings challenge both the "fecal-urethral" pathway and the "prevalence" hypotheses for the pathogenesis of FUTI in men and instead support alternative routes of infection in some men and the "special pathogenicity" hypothesis.
The fact that only a minority of subjects had the urine clone detected within the pretherapy rectal flora has several possible explanations. Conceivably, the urine clone actually might have been present as a minor component of the fecal population but was missed because of incomplete sampling (36). With only three colonies analyzed per plate, precise estimates of clonal contribution to the mixed E. coli population are impossible (the 95% confidence interval for a proportion of 0/3 ranges from 0% to 70%). Furthermore, a rectal swab samples only a minute fraction of the total intestinal E. coli population. Enhanced detection of minor fecal clones could be accomplished in future studies by intensified sampling, using hemadsorption to capture adhesin-expressing clones (11, 18), and/or using PCR to screen a mixed population for VF-positive strains (11, 18).
Alternatively, the urine clone may truly have been absent from the rectal flora at the time of the index UTI episode. This could occur if it had been introduced into the urinary tract at an earlier time, when it actually was a fecal colonizer, followed by asymptomatic residence within the bladder, upper urinary tract, or prostate but disappearance from the gut reservoir. This situation also could occur from de novo introduction of the clone into the host's urinary tract from an exogenous reservoir, including possibly a sex partner's vagina or rectum, without the clone ever residing in the host's gut. The former scenario is consistent with the known role of the prostate as a source for recurrent UTI episodes in men (23, 24), the latter with the occurrence of presumed sexual transmission of UTI-causing E. coli among both heterosexual and homosexual men (2, 3, 10, 43). Sexual transmission could be more directly assessed in future studies by comparing the subjects' urine and rectal isolates with concurrent rectal and vaginal isolates from sex partners (8).
Several findings supported the "special pathogenicity" hypothesis over the "prevalence" hypothesis as an explanation for why certain E. coli strains cause FUTI in men. First, the urine clone clearly was not highly prevalent within the fecal flora for most subjects. These included the 54% of subjects in whom it was not detected at all in the rectal culture (group 3), plus, probably, some proportion of the 22% of subjects in whom it was recovered from the rectal sample admixed with other clones (group 2). Second, even those urine clones that did predominate within the host's rectal flora, including to the extent of being the only detectable rectal clone (group 1), did not differ appreciably from other urine clones, including specifically those that were not detected at all within the rectal flora (group 3). Assuming that the putative and proven VFs included in our screening do predict virulence capability, as suggested by previous experimental data (14), this finding suggests that fecal predominance does not enable clones that lack "special pathogenicity" to cause FUTI in men, through a mass action-like effect. Third, the urine isolates collectively differed enormously from the unique rectal clones with respect to their overwhelmingly group B2 phylogenetic background, extensive virulence profiles, and frequent expression of virulence-associated surface antigens. The most parsimonious explanation for these findings, taken together, is that special pathogenicity is the main causal factor in FUTI in men, with prevalence contributing minimally if at all.
The striking differences between urine and rectal isolates also helped to identify specific traits that may contribute directly to pathogenesis or are linked to such traits. Interestingly, the stepwise multivariable logistic regression analysis suggested, based on order of entry of predictor variables into the model, that phylogenetic group B2 was the strongest overall predictor of urine source, ahead of all the individual VFs and papA alleles. This finding might simply indicate that group B2 status is a marker for the multiple group B2-associated VFs analyzed and so serves as a composite surrogate for them and benefits from their collective predictive power (12). Alternatively, particularly since the OR for group B2 did not change substantially as other (even group B2-associated) predictor variables were entered into the models, it may be that there are important group B2-associated VFs that were not included in our screen and do not strictly cosegregate with the studied VFs. This is highly plausible, given the large number of novel putative virulence genes that are being discovered currently through sequence analysis of group B2-derived pathogens (7, 42).
Multiple specific VFs significantly predicted the urine source according to one or more of the multivariate models. Although several are traditionally recognized and have been studied specifically in connection with prostatitis (e.g., hly and cnf) (1, 25, 26, 31, 32, 37), others are more recently discovered and/or have received comparatively little attention to date (e.g., ompT, malX, and iroN) (12, 15, 21, 34). Conceivably some of these or other genetically linked traits could serve as targets for preventive interventions if a direct contribution to pathogenesis can be confirmed, as has been done for IroN (34). Notably, that such linkages are important to consider was suggested by the marked sensitivity of the statistical models to small changes in the predictor variable pool, evidence that many variables were competing for almost the same variance and hence could substitute readily for one another depending on the specific analytical parameters.
An important strength of the study is the matched-control design, in which rectal isolates were obtained from the same hosts as the urine isolates, thereby controlling for multiple potential confounding factors and allowing essentially a by-host clonal comparison of the urine and rectal isolates (13, 30). In this respect, the present study supercedes a recent analysis in which an unrelated fecal control group was compared with these FUTI isolates (17). Additional strengths include the extensive range of bacterial traits analyzed, which included diverse virulence-associated genes, surface antigens, and phylogenetic background; the use of genomic profiling to resolve discrete clones; and the incorporation of host-specific data. Limitations include the incomplete sampling of fecal (and, conceivably, urinary) clonal diversity, possibly leading to false conclusions regarding the absence of the urine clone from the host's gut. Additionally, the strains derived from a single locale and time interval, possible limiting generalizability. Likewise, virulence was inferred based on surrogate markers and clinical behavior, not experimentation.
In summary, we found that in only a minority of men with FUTI did the urine clone predominate within the host's fecal flora upon presentation, and those urine clones that did so predominate were essentially indistinguishable from those that did not. In contrast, urine and rectal clones differed dramatically with respect to phylogenetic background, virulence gene content, and surface antigens, in patterns suggesting much greater virulence potential for the urine isolates. These findings challenge both the "fecal-urethral" pathway and the "prevalence" hypotheses for the pathogenesis of FUTI in men and instead strongly support the "special pathogenicity" hypothesis and, in some men, alternative infection routes or exacerbations of chronic bacterial prostatitis. They also identify specific bacterial traits as potential targets for protective interventions against FUTI.
ACKNOWLEDGMENTS
This material is based upon work supported by Office of Research and Development, Medical Research Service, Department of Veterans Affairs (J.R.J.), National Research Initiative (NRI) Competitive Grants Program/U.S. Department of Agriculture grant 00-35212-9408 (J.R.J.), the Scandinavian Society for Antimicrobial Chemotherapy (T.S.), and the Medical Society of Gteborg (P.U.).
Ann Emery assisted with manuscript preparation. Dave Prentiss prepared the figure.
REFERENCES
Andreu, A., A. E. Stapleton, C. Fennell, H. A. Lockman, M. Xercavins, F. Fernandez, and W. E. Stamm. 1997. Urovirulence determinants in Escherichia coli strains causing prostatitis. J. Infect. Dis. 176:464-469.
Bailey, R. R., B. A. Peddie, C. P. Swainson, and D. Kirkpatrick. 1986. Sexual acquisition of urinary tract infection in a man. Nephron 44:217-218.
Barnes, R. C., R. E. Roddy, R. Daifuku, and W. E. Stamm. 1986. Urinary-tract infection in sexually active homosexual men. Lancet i:171-173.
Berg, D. E., N. S. Akopyants, and D. Kersulyte. 1994. Fingerprinting microbial genomes using the RAPD or AP-PCR method. Methods Mol. Cell Biol. 5:13-24.
Centers for Disease Control and Prevention. 1996. Standardized molecular subtyping of Escherichia coli by pulsed-field gel electrophoresis. Training Course. Centers for Disease Control and Prevention, Atlanta, Ga.
Clermont, O., S. Bonacorsi, and E. Bingen. 2000. Rapid and simple determination of the Escherichia coli phylogenetic group. Appl. Environ. Microbiol. 66:4555-4558.
Dobrindt, U., G. Blum-Oehler, T. Hartsch, G. Gottschalk, E. Z. Ron, R. Funfstuck, and J. Hacker. 2001. S-fimbria-encoding determinant sfaI is located on pathogenicity island III536 of uropathogenic Escherichia coli strain 536. Infect. Immun. 69:4248-4256.
Foxman, B., S. D. Manning, P. Tallman, R. Bauer, L. Zhang, J. S. Koopman, B. Gillespie, J. D. Sobel, and C. F. Marrs. 2002. Uropathogenic Escherichia coli are more likely than commensal E. coli to be shared between heterosexual sex partners. Am. J. Epidemiol. 156:1133-1140.
Grüneberg, R. N. 1969. Relationship of infecting urinary organism to the faecal flora in patients with symptomatic urinary infection. Lancet i:766-768.
Hebelka, M., K. Lincoln, and T. Sandberg. 1993. Sexual acquisition of acute pyelonephritis in a man. Scand. J. Infect. Dis. 25:141-143.
Johnson, J. R., and J. J. Brown. 1998. Colonization with and acquisition of uropathogenic Escherichia coli strains as revealed by polymerase chain reaction-based detection. J. Infect. Dis. 177:1120-1124.
Johnson, J. R., P. Delavari, M. Kuskowski, and A. L. Stell. 2001. Phylogenetic distribution of extraintestinal virulence-associated traits in Escherichia coli. J. Infect. Dis. 183:78-88.
Johnson, J. R., N. Kaster, M. A. Kuskowski, and G. V. Ling. 2003. Identification of urovirulence traits in Escherichia coli by comparison of urinary and rectal E. coli isolates from dogs with urinary tract infection. J. Clin. Microbiol. 41:337-345.
Johnson, J. R., M. Kuskowski, E. Denamur, J. Elion, and B. Picard. 2000. Clonal origin, virulence factors, and virulence (letter and reply). Infect. Immun. 68:424-425.
Johnson, J. R., T. T. O'Bryan, M. A. Kuskowski, and J. N. Maslow. 2001. Ongoing horizontal and vertical transmission of virulence genes and papA alleles among Escherichia coli blood isolates from patients with diverse-source bacteremia. Infect. Immun. 69:5363-5374.
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., F. Scheutz, P. Ulleryd, M. A. Kuskowski, T. T. O'Bryan, and T. Sandberg. 2005. Host-pathogen relationships among Escherichia coli isolates from men with febrile urinary tract infection. Clin. Infect. Dis. 40:813-822.
Johnson, J. R., A. Stell, and P. Delavari. 2001. Canine feces as a reservoir of extraintestinal pathogenic Escherichia coli. Infect. Immun. 69:1306-1314.
Johnson, J. R., and A. L. Stell. 2000. Extended virulence genotypes of Escherichia coli strains from patients with urosepsis in relation to phylogeny and host compromise. J. Infect. Dis. 181:261-272.
Johnson, J. R., A. L. Stell, F. Scheutz, T. T. O'Bryan, T. A. Russo, U. B. Carlino, C. C. Fasching, J. Kavle, L. van Dijk, and W. Gaastra. 2000. Analysis of F antigen-specific papA alleles of extraintestinal pathogenic Escherichia coli using a novel multiplex PCR-based assay. Infect. Immun. 68:1587-1599.
Kanamura, S., H. Kurazono, S. Ishitoya, A. Terai, T. Habuchi, M. Nakano, O. Ogawa, and S. Yamamoto. 2003. Distribution and genetic association of putative uropathogenic virulence factors iroN, iha, kpsMT, ompT and usp in Escherichia coli isolated from urinary tract infections in Japan. J. Urol. 170:2490-2493.
Lidin-Janson, G., B. Kaijser, K. Lincoln, S. Olling, and H. Wedel. 1978. The homogeneity of the faecal coliform flora of normal school-girls, characterized by serological and biochemical properties. Med. Microbiol. Immunol. 164:247-253.
Lipsky, B. A. 1999. Prostatitis and urinary tract infection in men: what's new; what's true Am. J. Med. 106:327-334.
Lipsky, B. A. 1989. Urinary tract infections in men: epidemiology, pathophysiology, diagnosis, and treatment. Ann. Intern. Med. 110:138-150.
Mitsumori, K., A. Terai, S. Yamamoto, S. Ishitoya, and O. Yoshida. 1999. Virulence characteristics of Escherichia coli in acute bacterial prostatitis. J. Infect. Dis. 180:1378-1381.
O'Hanley, P., G. Lalonde, and G. Ji. 1991. -Hemolysin contributes to the pathogenicity of piliated digalactoside-binding Escherichia coli in the kidney: efficacy of an -hemolysin vaccine in preventing renal injury in the BALB/c mouse model of pyelonephritis. Infect. Immun. 59:1153-1161.
O'Hanley, P., D. Low, I. Romero, D. Lark, K. Vosti, S. Falkow, and G. Schoolnik. 1985. Gal-Gal binding and hemolysin phenotypes and genotypes associated with uropathogenic Escherichia coli. N. Engl. J. Med. 7:414-420.
Orskov, F., and I. Orskov. 1984. Serotyping of Escherichia coli. Methods Microbiol. 14:43-111.
Orskov, I., F. Orskov, A. Birch-Andersen, M. Kanamori, and C. Svanborg Eden. 1982. O, K, H and fimbrial antigens in Escherichia coli serotypes associated with pyelonephritis and cystitis. Scand. J. Infect. Dis. 33:18-25.
Plos, K., H. Connell, U. Jodal, B. I. Marklund, S. Mrild, B. Wettergren, and C. Svanborg. 1995. Intestinal carriage of P fimbriated Escherichia coli and the susceptibility to urinary tract infection in young children. J. Infect. Dis. 171:625-631.
Rippere-Lampe, K. E., M. Lang, H. Ceri, M. Olson, H. A. Lockman, and A. O'Brien. 2001. Cytotoxic necrotizing factor type 1-positive Escherichia coli causes increased inflammation and tissue damage to the prostate in a rat prostatitis model. Infect. Immun. 69:5615-6519.
Ruiz, J., K. Simon, J. P. Horcajada, M. Velasco, M. Barranco, G. Roig, A. Moreno-Martinez, J. A. Martinez, T. Jimenez de Anta, J. Mensa, and J. Vila. 2002. Differences in virulence factors among clinical isolates of Escherichia coli causing cystitis and pyelonephritis in women and prostatitis in men. J. Clin. Microbiol. 40:4445-4449.
Russo, T. A., and J. R. Johnson. 2003. Medical and economic impact of extraintestinal infections due to Escherichia coli: an overlooked epidemic. Microbes Infect. 5:449-456.
Russo, T. A., C. D. McFadden, U. B. Carlino-MacDonald, J. M. Beanan, T. J. Barnard, and J. R. Johnson. 2002. IroN functions as a siderophore receptor and is a urovirulence factor in an extraintestinal pathogenic isolate of Escherichia coli. Infect. Immun. 70:7156-7160.
Tenover, F. C., R. D. Arbeit, R. V. Goering, P. A. Mickelsen, B. E. Muray, D. H. Persing, and B. Swaminathan. 1995. Interpreting chromosomal DNA restriction patterns produced by pulsed-field gel electrophoresis: criteria for bacterial strain typing. J. Clin. Microbiol. 33:2233-2239.
Terai, A., S. Ishitoya, K. Mitsumori, and O. Ogawa. 2000. Molecular epidemiological evidence for ascending urethral infection in acute bacterial prostatitis. J. Urol. 164:1945-1947.
Terai, A., S. Yamamoto, K. Mitsumori, Y. Okada, H. Kurazono, Y. Takeda, and O. Yoshida. 1997. Escherichia coli virulence factors and serotypes in acute bacterial prostatitis. Int. J. Urol. 4:289-294.
Turck, M., R. G. Petersdorf, and M. R. Fournier. 1962. The epidemiology of nonenteric Escherichia coli infections: prevalence of serological groups. J. Clin. Investig. 41:1760-1765.
Ulleryd, P., K. Lincoln, F. Scheutz, and T. Sandberg. 1994. Virulence characteristics of Escherichia coli in relation to host response in men with symptomatic urinary tract infection. Clin. Infect. Dis. 18:579-584.
Ulleryd, P., B. Zachrisson, G. Aus, S. Bergdahl, J. Hugosson, and T. Sandberg. 2001. Selective urological evaluation in men with febrile urinary tract infection. BJU Int. 88:15-20.
Ulleryd, P., B. Zackrisson, G. Aus, S. Bergdahl, J. Hugosson, and T. Sandberg. 1999. Prostatic involvement in men with febrile urinary tract infection as measured by serum prostate-specific antigen and transrectal ultrasonography. BJU Int. 84:470-474.
Welch, R., V. Burland, G. R. Plunkett III, P. Redford, P. Roesch, D. Rasko, E. L. Buckles, S. R. Liou, A. Boutin, J. Hackett, D. Stroud, G. F. Mayhew, D. J. Rose, S. Zhou, D. C. Schwartz, N. T. Perna, H. L. Mobley, M. S. Donnenberg, and F. R. Blattner. 2002. Extensive mosaic structure revealed by the complete genome sequence of uropathogenic Escherichia coli. Proc. Natl. Acad. Sci. USA 99:17020-17024.
Wong, E. S., and W. E. Stamm. 1983. Sexual acquisition of urinary tract infection in a man. JAMA 250:3087-3088.
Yamamoto, S., T. Tsukamoto, A. Terai, H. Kurazono, Y. Takeda, and O. Yoshida. 1997. Genetic evidence supporting the fecal-perineal-urethral hypothesis in cystitis caused by Escherichia coli. J. Urol. 157:1127-1129.(James R. Johnson, Flemmin)
Geriatric Research, Education, and Clinical Center, Minneapolis VA Medical Center
Departments of Medicine
Psychiatry, University of Minnesota, Minneapolis, Minnesota
International Escherichia coli and Klebsiella Center, Statens Seruminstitut, Copenhagen, Denmark
Department of Infectious Diseases, University of Gteborg, Gteborg, Sweden
ABSTRACT
Among men with febrile urinary tract infection (FUTI), whether the host's fecal flora is the source for the urine strain ("fecal-urethral" hypothesis), and whether pathogenesis is driven by prevalence versus special pathogenicity, are unknown. Accordingly, pretherapy urine isolates from 65 men with FUTI were compared with concurrent rectal isolates from the same hosts according to serotype, genomic profile, phylogenetic group, and virulence genotype. The host's multiple rectal colonies included only the urine clone in 25% of subjects, the urine clone plus additional clones in 22%, and only nonurine clones in 54%. Compared with the 67 unique rectal clones, the 65 urine isolates were significantly enriched for phylogenetic group B2, virulence-associated serotypes, and specific virulence genes and contained more virulence genes (median, 10 versus 6: P < 0.001). In multivariable models, phylogenetic group B2, hlyD (hemolysin), cnf1 (cytotoxic necrotizing factor), iroN (siderophore receptor), ompT (outer membrane protease), and malX (pathogenicity island marker) most strongly predicted urine source. These findings challenge the fecal-urethral and prevalence hypotheses for FUTI pathogenesis and instead strongly support the possibility of alternate infection routes in some men and the special pathogenicity hypothesis. They also identify specific bacterial traits as potential targets for anti-FUTI interventions.
INTRODUCTION
Escherichia coli is a major extraintestinal pathogen, causing tremendous morbidity, mortality, and increased health care costs (33). Urinary tract infection (UTI), the principal extraintestinal syndrome caused by E. coli, although most common among women, is also a significant problem among men, who can experience cystitis, pyelonephritis, acute and chronic prostatitis, and febrile UTI (FUTI) (23, 24, 41). Better understanding of the pathogenesis of UTI in men is needed to guide the development of effective preventive measures.
It is unknown whether in men with UTI the causative E. coli strains usually derive immediately from the host's own intestinal flora, as is true in females, according to the "fecal-urethral" hypothesis (9, 44). For example, vaginal or anal intercourse may introduce microorganisms directly into a man's urethra from his sex partner's vaginal or rectal flora, without involving his intestinal tract (2, 3, 8, 43). Likewise, it is not known whether the particular E. coli strains that cause UTI in men do so merely because of their high prevalence within the host's intestinal flora, consistent with the "prevalence" hypothesis for UTI pathogenesis (38), or instead because they possess an enhanced ability to cause extraintestinal disease compared with other E. coli strains, consistent with the "special pathogenicity" hypothesis that is thought to underlie UTI pathogenesis in most females (27, 30). Although the virulence factors (VF) and surface antigens of E. coli isolates from men with febrile UTI and prostatitis exhibit many similarities to those of strains causing pyelonephritis in women and girls, consistent with the "special pathogenicity" hypothesis (1, 25, 32, 37, 39), few data are available comparing male UTI isolates with fecal isolates from the same hosts (36).
Accordingly, we compared a collection of E. coli urine isolates from a cohort of 65 men with FUTI (a subset of the 70 isolates described in a recent report [17]) with concurrent pretherapy fecal E. coli isolates from the same men to assess both the "fecal-urethral" pathway and the "prevalence" versus "special pathogenicity" hypotheses in the context of FUTI in men. Specifically, we used genomic profiling to resolve distinct clones among the subjects' multiple fecal isolates and then compared urine and fecal clones with respect to surface antigens, phylogenetic background, and VF profiles.
MATERIALS AND METHODS
Subjects and isolates. The 65 men with FUTI due to E. coli were prospectively enrolled at Sahlgrenska University Hospital, Gteborg, Sweden, between March 1993 and December 1996, as reported in greater detail elsewhere (17). Briefly, to be included, patients had to have a fever of 38.0°C and at least one symptom or sign referable to the urinary tract (frequency, dysuria, flank pain, or costovertebral angle tenderness), as well as 104 CFU/ml of E. coli in midstream urine. No patient presented with classical acute prostatitis, although eight had slight prostatic tenderness. None developed symptoms while in hospital. All patients were hospitalized for at least 1 day. Patients with a chronic indwelling urinary catheter were excluded. No patient had known human immunodeficiency virus infection. The University of Gteborg Research Ethics Committee approved the study protocol. All subjects provided informed consent to participate.
Signs and symptoms, a history of previous UTI and genitourinary disorders, concomitant diseases, and a clinical evaluation of the patient's general condition were recorded. Urine, blood samples, and a rectal swab for culture were obtained before antimicrobial treatment. The serum prostate-specific antigen (PSA) level was measured (normal, 4.0 ng/ml) (41). Uroflowmetry, digital rectal examination, and measurement of postvoid residual urine by abdominal ultrasound were done at entry and after 3 months. Cystourethroscopy and imaging studies of the upper urinary tract were performed about 3 months after enrollment, unless carried out within the last 5 years (40).
Urine and rectal cultures. Urine was cultured semiquantitatively on blood agar and cysteine-lactose-electrolyte-deficient agar plates using a calibrated loop technique. A single colony was saved for further analysis. Rectal swabs were cultured on modified Drigalski agar plates (22). From each plate, the last three colonies at the end of the streak area, plus any other morphologically distinct colonies, were saved for further analysis. This method provides 97% probability of capturing the quantitatively predominant clone in the sample (22). The 65 subjects described here represent all of the previously described 70 men with E. coli FUTI (17) from whom rectal E. coli isolates were available for analysis. (For the remaining five subjects, the initial rectal isolates either were not saved or were not retrievable from storage.)
Selection and clonal analysis of urine and rectal isolates. Each host's urine isolate and multiple (3) fecal isolates underwent O:K:H (somatic, capsule, and flagellar) antigen detection, according to standard methods (28). For the fecal isolates from 10 subjects, a truncated serotyping approach was used in which K and H antigens were determined only if there was an O-antigen match with the host's urine isolate.
To resolve discrete clones, all isolates (urine and rectal) from a given host were compared according to genomic profiles as generated using random amplified polymorphic DNA (RAPD) analysis. Random oligomer 1247 was used, supplemented in some instances with oligomer 1290, with visual inspection of profiles in ethidium bromide-stained agarose gels used to define unique clones (4). Serologically similar isolates that exhibited indistinguishable RAPD profiles were defined as representing the same clone, whereas serologically distinct isolates that exhibited distinct RAPD profiles were defined as representing different clones. Ambiguous results, and any discrepancies between the serological data and RAPD analysis, were further assessed by pulsed-field gel electrophoresis (5), with a three-band difference used as the threshold criterion for defining clonality (35).
Phylotyping and virulence genotyping. All urine isolates and at least one representative of each unique clone among each host's rectal isolates underwent PCR-based testing for major E. coli phylogenetic group (A, B1, B2, and D) (6), extended virulence genotype (for 34 markers) (19), and, if positive for any pap (P fimbriae) element, 12 alleles of papA (P fimbriae structural subunit) (20). All assays were done at least in duplicate using independently prepared lysates of the test strains and appropriate positive and negative controls. The VF score was the sum of virulence markers detected in an isolate, adjusted for multiple detection of the pap, sfa/foc (S/F1C fimbriae), and kps II (group 2 capsule) operons.
Statistical methods. Hosts and the corresponding urine isolates were stratified for analysis into three groups, according to the degree of commonality among the host's urine and rectal clones. Group 1 comprised hosts for whom all rectal isolates represented the urine clone, group 2 those whose rectal isolates included the urine clone but also at least one nonurine clone, and group 3 those with only nonurine rectal clones. Host characteristics and acute manifestations of FUTI that were analyzed included bacteremia, flank pain, acute PSA level (absolute and categorically as >4 ng/ml), host age (absolute and categorically as 45, 50, and >65), previous UTI, previous FUTI, increased postvoid residual volume, recent urinary tract instrumentation, renal cortical scarring, diabetes mellitus, and a combination of (age >65, elevated PSA level, and 1 possibly predisposing condition), as previously described in greater detail (17). For analytical purposes, 10 UTI-associated O serogroups, i.e., O1, O2, O4, O6, O7, O8, O16, O18, O25, and O75, were defined as the O-UTI groups (16, 29), six UTI-associated K antigens, i.e., K1, K2, K3, K5, K12, and K13, were defined as the K-UTI types (16), and 12 pyelonephritis-associated O:K:H serotypes, i.e., O1:K1:H7, O2:K1:H4, O4:K5:H5, O4:K12:H1, O4:K12:H5, O6:K2:H1, O6:K5:H1, O7:K1:H1, O16:K1:H6, O18:K1:H7, O18:K5:H–, and O18:K5:H7, were defined as the O:K:H-Pyelo serotypes (16, 29). Isolates conforming to one of the O:K:H-Pyelo serotypes except for an absent K or H antigen were analyzed as belonging to that serotype (16).
Comparisons of proportions were tested using Fisher's exact test (two-tailed). Comparisons of continuous variables were tested using the Mann-Whitney U test (exact). Odds ratios (ORs), with 95% confidence intervals (CIs), for individual or multiple variables as predictors of urine versus rectal origin were calculated using, respectively, univariate or stepwise multiple logistic regression analysis. Because of multiple comparisons, the significance criterion was a P value of 0.01.
RESULTS
Clonal analysis of urine and rectal isolates. According to O:K:H serotyping, RAPD analysis, and selective pulsed-field gel electrophoresis analysis of concurrent pretherapy urine and rectal isolates from 65 men with FUTI, for 16 (25%) of the subjects (group 1), the host's urine clone accounted for all of the multiple rectal isolates examined. In contrast, for 49 (75%) of the subjects, the rectal isolates included one or more distinct (nonurine) clones, either together with the host's urine clone (group 2; n = 14 [22%]) or in the absence of the urine clone (group 3; n = 35 [54%]). Thus, the urine clone clearly predominated within the host's rectal flora in 25% of subjects (group 1), clearly did not predominate there in 54% of subjects (group 3), and may or may not have predominated in 22% of subjects (group 2). Overall, the host's urine clone was identified within the host's rectal flora for 30 (46%) of the 65 subjects (groups 1 and 2). The 49 subjects (76%) from groups 2 and 3 each had from 1 to 3 unique (nonurine) rectal clones identified, giving a total of 67 unique rectal clones for comparison with the 65 urine clones.
Urine strain characteristics versus host colonization status. If high prevalence within the rectal flora could allow a less-virulent clone to cause FUTI, then the host's rectal colonization status with respect to the urine clone (versus other E. coli clones) should influence the characteristics of the urine clone. Therefore, comparisons were made among the 65 FUTI urine isolates according to 53 bacterial characteristics (serological categories, phylogenetic groups, and virulence genes) in relation to rectal colonization status. In comparisons of group 1 versus groups 2 and 3 combined, group 3 versus groups 1 and 2 combined, and group 1 versus group 3, the resulting 159 comparisons yielded no statistically significant differences and only two borderline significant differences (for both, P = 0.03), which could have arisen by chance alone. This provided evidence against prevalence within the rectal flora as a significant determinant of bacterial characteristics in FUTI and supported combining the urine isolates as an "all urine" group.
Host characteristics and acute manifestations according to rectal colonization status. Analogous to the minimal differences noted according to bacterial characteristics, minimal differences also were detected among the 3 host subgroups when they were compared for host characteristics and acute manifestations of FUTI (16 variables total), using the same groupings as described above. Only 1 of the 48 resulting comparisons yielded a statistically significant association, and 2 yielded borderline significant associations, all involving age. That is, group 3 hosts (who had only nonurine rectal clones) were somewhat less likely to be >65 years old (8/35; 23%) than were either group 1 hosts (9/16; 56%; P = 0.03) or all other hosts combined (17/30; 57%; P = 0.01) and were somewhat younger in absolute terms (median age, 55 years [18 to 86] than other hosts (median age, 69 years [32 to 85] (P = 0.025). Since these associations were not particularly strong and, in view of the multiple comparisons, could have arisen by chance alone, the three subgroups of urine isolates were combined as an "all urine" group for enhanced statistical power in comparisons versus the unique rectal isolates.
Urine isolates versus unique rectal strains. Compared with the 67 unique rectal strains, the 65 urine isolates were strikingly enriched for phylogenetic group B2 and most of the studied VFs and virulence-associated serological categories (Table 1). Accordingly, aggregate VF scores were substantially higher among the urine isolates than among the rectal isolates (median score, 10 versus 6: P < 0.001) (Fig. 1). Likewise, among the 12 papA alleles the urine isolates exhibited a significantly greater prevalence of the F10 variant (25% versus 6%: P = 0.002) and similar borderline trends for the F16 and F48 variants (P = 0.03 for both). Similar results, albeit with slightly higher P values due to the smaller sample size, were obtained when, for optimal matching, the analysis was limited to those urine isolates from the 49 hosts (groups 2 and 3) who contributed 1 unique rectal clone (not shown).
Because of the multiplicity of traits associated with urine origin by univariate analysis, stepwise multiple logistic regression analysis was used to identify independent predictors of urine source. A series of models was constructed using different groups of molecularly defined bacterial traits as candidate predictor variables (Table 2). The variables that in one or more of these models significantly predicted urine source included phylogenetic group B2 (the strongest predictor overall), hlyD, ompT, malX, and cnf1 (Table 2). In contrast, in two models the F12 papA allele was a significant negative predictor of urine source (Table 2).
These analyses were repeated with the three serological categories (i.e., O-UTI, K-UTI, and O:K:H-pyelo) as added candidate predictor variables, which reduced the sample size from 132 to 117 because of incomplete serological data for 15 fecal isolates. The results were largely unchanged overall, and no serological category emerged as a significant predictor in any model (not shown). Differences compared with Table 2 included, in model 1, a step 4 in which iroN emerged as another significant positive predictor of urine source (OR, 7.6 [CI, 2.0 to 29.4]; P = 0.003) and a step 5 in which ibeA also emerged as a significant negative predictor (OR, 0.1 [CI, 0.02 to 0.6]; P = 0.009). Additionally, in models 2 and 3, iroN (OR, 13.0 [CI, 3.6 to 46.1] in model 2, and OR, 6.5 [CI 2.5 to 17.4] in model 3; P < 0.001 for both) replaced ompT as a positive predictor, whereas model 2 stopped after step 2 (not shown).
To clarify the basis for the strong predictive power of phylogenetic group B2, the phylogenetic distribution of VFs was assessed among the combined urine and rectal isolates (n = 132). Nearly all of the VFs that were significantly associated with urine source (Table 1) were also highly associated with group B2 (not shown). Accordingly, VF scores were much higher for group B2 than for the other phylogenetic groups combined (median, 10 [range, 2 to 15], versus. 3.5 [range, 0 to 12]; P < 0.001) (Fig. 1). This relationship held independently of urine versus rectal source. That is, urine and rectal source group B2 isolates exhibited similarly high median VF scores (10 in both groups), although the range extended lower among the rectal isolates (urine, [8 to 14]; rectal, [2 to 15]; P = 0.04). Likewise, urine and rectal source non-B2 isolates exhibited similarly low VF scores (median, 3.0 [1 to 8] versus 3.5 [0 to 12]; P > 0.10).
DISCUSSION
In this molecular epidemiological and phylogenetic comparison of urine and concurrent rectal E. coli isolates from men with FUTI, we found that the host's urine clone predominated within the host's gut flora at the time of presentation in only a minority of subjects. Moreover, those urine clones that did predominate rectally were essentially indistinguishable from those that did not, as were the corresponding hosts, except for possible minor age differences. In contrast, urine clones collectively differed dramatically from rectal clones with respect to phylogenetic background, virulence gene content, and serotypes, in patterns suggesting much greater virulence potential for the urine isolates. These findings challenge both the "fecal-urethral" pathway and the "prevalence" hypotheses for the pathogenesis of FUTI in men and instead support alternative routes of infection in some men and the "special pathogenicity" hypothesis.
The fact that only a minority of subjects had the urine clone detected within the pretherapy rectal flora has several possible explanations. Conceivably, the urine clone actually might have been present as a minor component of the fecal population but was missed because of incomplete sampling (36). With only three colonies analyzed per plate, precise estimates of clonal contribution to the mixed E. coli population are impossible (the 95% confidence interval for a proportion of 0/3 ranges from 0% to 70%). Furthermore, a rectal swab samples only a minute fraction of the total intestinal E. coli population. Enhanced detection of minor fecal clones could be accomplished in future studies by intensified sampling, using hemadsorption to capture adhesin-expressing clones (11, 18), and/or using PCR to screen a mixed population for VF-positive strains (11, 18).
Alternatively, the urine clone may truly have been absent from the rectal flora at the time of the index UTI episode. This could occur if it had been introduced into the urinary tract at an earlier time, when it actually was a fecal colonizer, followed by asymptomatic residence within the bladder, upper urinary tract, or prostate but disappearance from the gut reservoir. This situation also could occur from de novo introduction of the clone into the host's urinary tract from an exogenous reservoir, including possibly a sex partner's vagina or rectum, without the clone ever residing in the host's gut. The former scenario is consistent with the known role of the prostate as a source for recurrent UTI episodes in men (23, 24), the latter with the occurrence of presumed sexual transmission of UTI-causing E. coli among both heterosexual and homosexual men (2, 3, 10, 43). Sexual transmission could be more directly assessed in future studies by comparing the subjects' urine and rectal isolates with concurrent rectal and vaginal isolates from sex partners (8).
Several findings supported the "special pathogenicity" hypothesis over the "prevalence" hypothesis as an explanation for why certain E. coli strains cause FUTI in men. First, the urine clone clearly was not highly prevalent within the fecal flora for most subjects. These included the 54% of subjects in whom it was not detected at all in the rectal culture (group 3), plus, probably, some proportion of the 22% of subjects in whom it was recovered from the rectal sample admixed with other clones (group 2). Second, even those urine clones that did predominate within the host's rectal flora, including to the extent of being the only detectable rectal clone (group 1), did not differ appreciably from other urine clones, including specifically those that were not detected at all within the rectal flora (group 3). Assuming that the putative and proven VFs included in our screening do predict virulence capability, as suggested by previous experimental data (14), this finding suggests that fecal predominance does not enable clones that lack "special pathogenicity" to cause FUTI in men, through a mass action-like effect. Third, the urine isolates collectively differed enormously from the unique rectal clones with respect to their overwhelmingly group B2 phylogenetic background, extensive virulence profiles, and frequent expression of virulence-associated surface antigens. The most parsimonious explanation for these findings, taken together, is that special pathogenicity is the main causal factor in FUTI in men, with prevalence contributing minimally if at all.
The striking differences between urine and rectal isolates also helped to identify specific traits that may contribute directly to pathogenesis or are linked to such traits. Interestingly, the stepwise multivariable logistic regression analysis suggested, based on order of entry of predictor variables into the model, that phylogenetic group B2 was the strongest overall predictor of urine source, ahead of all the individual VFs and papA alleles. This finding might simply indicate that group B2 status is a marker for the multiple group B2-associated VFs analyzed and so serves as a composite surrogate for them and benefits from their collective predictive power (12). Alternatively, particularly since the OR for group B2 did not change substantially as other (even group B2-associated) predictor variables were entered into the models, it may be that there are important group B2-associated VFs that were not included in our screen and do not strictly cosegregate with the studied VFs. This is highly plausible, given the large number of novel putative virulence genes that are being discovered currently through sequence analysis of group B2-derived pathogens (7, 42).
Multiple specific VFs significantly predicted the urine source according to one or more of the multivariate models. Although several are traditionally recognized and have been studied specifically in connection with prostatitis (e.g., hly and cnf) (1, 25, 26, 31, 32, 37), others are more recently discovered and/or have received comparatively little attention to date (e.g., ompT, malX, and iroN) (12, 15, 21, 34). Conceivably some of these or other genetically linked traits could serve as targets for preventive interventions if a direct contribution to pathogenesis can be confirmed, as has been done for IroN (34). Notably, that such linkages are important to consider was suggested by the marked sensitivity of the statistical models to small changes in the predictor variable pool, evidence that many variables were competing for almost the same variance and hence could substitute readily for one another depending on the specific analytical parameters.
An important strength of the study is the matched-control design, in which rectal isolates were obtained from the same hosts as the urine isolates, thereby controlling for multiple potential confounding factors and allowing essentially a by-host clonal comparison of the urine and rectal isolates (13, 30). In this respect, the present study supercedes a recent analysis in which an unrelated fecal control group was compared with these FUTI isolates (17). Additional strengths include the extensive range of bacterial traits analyzed, which included diverse virulence-associated genes, surface antigens, and phylogenetic background; the use of genomic profiling to resolve discrete clones; and the incorporation of host-specific data. Limitations include the incomplete sampling of fecal (and, conceivably, urinary) clonal diversity, possibly leading to false conclusions regarding the absence of the urine clone from the host's gut. Additionally, the strains derived from a single locale and time interval, possible limiting generalizability. Likewise, virulence was inferred based on surrogate markers and clinical behavior, not experimentation.
In summary, we found that in only a minority of men with FUTI did the urine clone predominate within the host's fecal flora upon presentation, and those urine clones that did so predominate were essentially indistinguishable from those that did not. In contrast, urine and rectal clones differed dramatically with respect to phylogenetic background, virulence gene content, and surface antigens, in patterns suggesting much greater virulence potential for the urine isolates. These findings challenge both the "fecal-urethral" pathway and the "prevalence" hypotheses for the pathogenesis of FUTI in men and instead strongly support the "special pathogenicity" hypothesis and, in some men, alternative infection routes or exacerbations of chronic bacterial prostatitis. They also identify specific bacterial traits as potential targets for protective interventions against FUTI.
ACKNOWLEDGMENTS
This material is based upon work supported by Office of Research and Development, Medical Research Service, Department of Veterans Affairs (J.R.J.), National Research Initiative (NRI) Competitive Grants Program/U.S. Department of Agriculture grant 00-35212-9408 (J.R.J.), the Scandinavian Society for Antimicrobial Chemotherapy (T.S.), and the Medical Society of Gteborg (P.U.).
Ann Emery assisted with manuscript preparation. Dave Prentiss prepared the figure.
REFERENCES
Andreu, A., A. E. Stapleton, C. Fennell, H. A. Lockman, M. Xercavins, F. Fernandez, and W. E. Stamm. 1997. Urovirulence determinants in Escherichia coli strains causing prostatitis. J. Infect. Dis. 176:464-469.
Bailey, R. R., B. A. Peddie, C. P. Swainson, and D. Kirkpatrick. 1986. Sexual acquisition of urinary tract infection in a man. Nephron 44:217-218.
Barnes, R. C., R. E. Roddy, R. Daifuku, and W. E. Stamm. 1986. Urinary-tract infection in sexually active homosexual men. Lancet i:171-173.
Berg, D. E., N. S. Akopyants, and D. Kersulyte. 1994. Fingerprinting microbial genomes using the RAPD or AP-PCR method. Methods Mol. Cell Biol. 5:13-24.
Centers for Disease Control and Prevention. 1996. Standardized molecular subtyping of Escherichia coli by pulsed-field gel electrophoresis. Training Course. Centers for Disease Control and Prevention, Atlanta, Ga.
Clermont, O., S. Bonacorsi, and E. Bingen. 2000. Rapid and simple determination of the Escherichia coli phylogenetic group. Appl. Environ. Microbiol. 66:4555-4558.
Dobrindt, U., G. Blum-Oehler, T. Hartsch, G. Gottschalk, E. Z. Ron, R. Funfstuck, and J. Hacker. 2001. S-fimbria-encoding determinant sfaI is located on pathogenicity island III536 of uropathogenic Escherichia coli strain 536. Infect. Immun. 69:4248-4256.
Foxman, B., S. D. Manning, P. Tallman, R. Bauer, L. Zhang, J. S. Koopman, B. Gillespie, J. D. Sobel, and C. F. Marrs. 2002. Uropathogenic Escherichia coli are more likely than commensal E. coli to be shared between heterosexual sex partners. Am. J. Epidemiol. 156:1133-1140.
Grüneberg, R. N. 1969. Relationship of infecting urinary organism to the faecal flora in patients with symptomatic urinary infection. Lancet i:766-768.
Hebelka, M., K. Lincoln, and T. Sandberg. 1993. Sexual acquisition of acute pyelonephritis in a man. Scand. J. Infect. Dis. 25:141-143.
Johnson, J. R., and J. J. Brown. 1998. Colonization with and acquisition of uropathogenic Escherichia coli strains as revealed by polymerase chain reaction-based detection. J. Infect. Dis. 177:1120-1124.
Johnson, J. R., P. Delavari, M. Kuskowski, and A. L. Stell. 2001. Phylogenetic distribution of extraintestinal virulence-associated traits in Escherichia coli. J. Infect. Dis. 183:78-88.
Johnson, J. R., N. Kaster, M. A. Kuskowski, and G. V. Ling. 2003. Identification of urovirulence traits in Escherichia coli by comparison of urinary and rectal E. coli isolates from dogs with urinary tract infection. J. Clin. Microbiol. 41:337-345.
Johnson, J. R., M. Kuskowski, E. Denamur, J. Elion, and B. Picard. 2000. Clonal origin, virulence factors, and virulence (letter and reply). Infect. Immun. 68:424-425.
Johnson, J. R., T. T. O'Bryan, M. A. Kuskowski, and J. N. Maslow. 2001. Ongoing horizontal and vertical transmission of virulence genes and papA alleles among Escherichia coli blood isolates from patients with diverse-source bacteremia. Infect. Immun. 69:5363-5374.
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., F. Scheutz, P. Ulleryd, M. A. Kuskowski, T. T. O'Bryan, and T. Sandberg. 2005. Host-pathogen relationships among Escherichia coli isolates from men with febrile urinary tract infection. Clin. Infect. Dis. 40:813-822.
Johnson, J. R., A. Stell, and P. Delavari. 2001. Canine feces as a reservoir of extraintestinal pathogenic Escherichia coli. Infect. Immun. 69:1306-1314.
Johnson, J. R., and A. L. Stell. 2000. Extended virulence genotypes of Escherichia coli strains from patients with urosepsis in relation to phylogeny and host compromise. J. Infect. Dis. 181:261-272.
Johnson, J. R., A. L. Stell, F. Scheutz, T. T. O'Bryan, T. A. Russo, U. B. Carlino, C. C. Fasching, J. Kavle, L. van Dijk, and W. Gaastra. 2000. Analysis of F antigen-specific papA alleles of extraintestinal pathogenic Escherichia coli using a novel multiplex PCR-based assay. Infect. Immun. 68:1587-1599.
Kanamura, S., H. Kurazono, S. Ishitoya, A. Terai, T. Habuchi, M. Nakano, O. Ogawa, and S. Yamamoto. 2003. Distribution and genetic association of putative uropathogenic virulence factors iroN, iha, kpsMT, ompT and usp in Escherichia coli isolated from urinary tract infections in Japan. J. Urol. 170:2490-2493.
Lidin-Janson, G., B. Kaijser, K. Lincoln, S. Olling, and H. Wedel. 1978. The homogeneity of the faecal coliform flora of normal school-girls, characterized by serological and biochemical properties. Med. Microbiol. Immunol. 164:247-253.
Lipsky, B. A. 1999. Prostatitis and urinary tract infection in men: what's new; what's true Am. J. Med. 106:327-334.
Lipsky, B. A. 1989. Urinary tract infections in men: epidemiology, pathophysiology, diagnosis, and treatment. Ann. Intern. Med. 110:138-150.
Mitsumori, K., A. Terai, S. Yamamoto, S. Ishitoya, and O. Yoshida. 1999. Virulence characteristics of Escherichia coli in acute bacterial prostatitis. J. Infect. Dis. 180:1378-1381.
O'Hanley, P., G. Lalonde, and G. Ji. 1991. -Hemolysin contributes to the pathogenicity of piliated digalactoside-binding Escherichia coli in the kidney: efficacy of an -hemolysin vaccine in preventing renal injury in the BALB/c mouse model of pyelonephritis. Infect. Immun. 59:1153-1161.
O'Hanley, P., D. Low, I. Romero, D. Lark, K. Vosti, S. Falkow, and G. Schoolnik. 1985. Gal-Gal binding and hemolysin phenotypes and genotypes associated with uropathogenic Escherichia coli. N. Engl. J. Med. 7:414-420.
Orskov, F., and I. Orskov. 1984. Serotyping of Escherichia coli. Methods Microbiol. 14:43-111.
Orskov, I., F. Orskov, A. Birch-Andersen, M. Kanamori, and C. Svanborg Eden. 1982. O, K, H and fimbrial antigens in Escherichia coli serotypes associated with pyelonephritis and cystitis. Scand. J. Infect. Dis. 33:18-25.
Plos, K., H. Connell, U. Jodal, B. I. Marklund, S. Mrild, B. Wettergren, and C. Svanborg. 1995. Intestinal carriage of P fimbriated Escherichia coli and the susceptibility to urinary tract infection in young children. J. Infect. Dis. 171:625-631.
Rippere-Lampe, K. E., M. Lang, H. Ceri, M. Olson, H. A. Lockman, and A. O'Brien. 2001. Cytotoxic necrotizing factor type 1-positive Escherichia coli causes increased inflammation and tissue damage to the prostate in a rat prostatitis model. Infect. Immun. 69:5615-6519.
Ruiz, J., K. Simon, J. P. Horcajada, M. Velasco, M. Barranco, G. Roig, A. Moreno-Martinez, J. A. Martinez, T. Jimenez de Anta, J. Mensa, and J. Vila. 2002. Differences in virulence factors among clinical isolates of Escherichia coli causing cystitis and pyelonephritis in women and prostatitis in men. J. Clin. Microbiol. 40:4445-4449.
Russo, T. A., and J. R. Johnson. 2003. Medical and economic impact of extraintestinal infections due to Escherichia coli: an overlooked epidemic. Microbes Infect. 5:449-456.
Russo, T. A., C. D. McFadden, U. B. Carlino-MacDonald, J. M. Beanan, T. J. Barnard, and J. R. Johnson. 2002. IroN functions as a siderophore receptor and is a urovirulence factor in an extraintestinal pathogenic isolate of Escherichia coli. Infect. Immun. 70:7156-7160.
Tenover, F. C., R. D. Arbeit, R. V. Goering, P. A. Mickelsen, B. E. Muray, D. H. Persing, and B. Swaminathan. 1995. Interpreting chromosomal DNA restriction patterns produced by pulsed-field gel electrophoresis: criteria for bacterial strain typing. J. Clin. Microbiol. 33:2233-2239.
Terai, A., S. Ishitoya, K. Mitsumori, and O. Ogawa. 2000. Molecular epidemiological evidence for ascending urethral infection in acute bacterial prostatitis. J. Urol. 164:1945-1947.
Terai, A., S. Yamamoto, K. Mitsumori, Y. Okada, H. Kurazono, Y. Takeda, and O. Yoshida. 1997. Escherichia coli virulence factors and serotypes in acute bacterial prostatitis. Int. J. Urol. 4:289-294.
Turck, M., R. G. Petersdorf, and M. R. Fournier. 1962. The epidemiology of nonenteric Escherichia coli infections: prevalence of serological groups. J. Clin. Investig. 41:1760-1765.
Ulleryd, P., K. Lincoln, F. Scheutz, and T. Sandberg. 1994. Virulence characteristics of Escherichia coli in relation to host response in men with symptomatic urinary tract infection. Clin. Infect. Dis. 18:579-584.
Ulleryd, P., B. Zachrisson, G. Aus, S. Bergdahl, J. Hugosson, and T. Sandberg. 2001. Selective urological evaluation in men with febrile urinary tract infection. BJU Int. 88:15-20.
Ulleryd, P., B. Zackrisson, G. Aus, S. Bergdahl, J. Hugosson, and T. Sandberg. 1999. Prostatic involvement in men with febrile urinary tract infection as measured by serum prostate-specific antigen and transrectal ultrasonography. BJU Int. 84:470-474.
Welch, R., V. Burland, G. R. Plunkett III, P. Redford, P. Roesch, D. Rasko, E. L. Buckles, S. R. Liou, A. Boutin, J. Hackett, D. Stroud, G. F. Mayhew, D. J. Rose, S. Zhou, D. C. Schwartz, N. T. Perna, H. L. Mobley, M. S. Donnenberg, and F. R. Blattner. 2002. Extensive mosaic structure revealed by the complete genome sequence of uropathogenic Escherichia coli. Proc. Natl. Acad. Sci. USA 99:17020-17024.
Wong, E. S., and W. E. Stamm. 1983. Sexual acquisition of urinary tract infection in a man. JAMA 250:3087-3088.
Yamamoto, S., T. Tsukamoto, A. Terai, H. Kurazono, Y. Takeda, and O. Yoshida. 1997. Genetic evidence supporting the fecal-perineal-urethral hypothesis in cystitis caused by Escherichia coli. J. Urol. 157:1127-1129.(James R. Johnson, Flemmin)