Pulsed-Field Gel Electrophoresis Typing of Escherichia coli Strains from Samples Collected before and after Pivmecillinam or Placebo Treatme
http://www.100md.com
微生物临床杂志 2006年第5期
National Center for Antimicrobials and Infection Control, Statens Serum Institut, Copenhagen, Denmark
Department of Clinical Microbiology, Hvidovre, Denmark
Department of Clinical Microbiology, Umea, Sweden
Umea Primary Health Care Centre, Umea, Sweden
Department of Medical Microbiology and Immunology, University of Gothenburg, Gothenburg, Sweden
ABSTRACT
The primary infecting Escherichia coli strains from 156 women with community-acquired uncomplicated urinary tract infection (UTI) randomized to pivmecillinam or placebo and the E. coli strains causing UTI at two follow-up visits were typed using pulsed-field gel electrophoresis (PFGE). In the pivmecillinam treatment group PFGE showed that among patients having a negative urine culture at the first follow-up 77% (46/60) had a relapse with the primary infecting E. coli strain and 23% (14/60) had reinfection with a new E. coli strain at the second follow-up. Among patients having E. coli at the first follow-up PFGE showed that 80% (32/40) had persistence with the primary infecting E. coli strain, 15% (6/40) had reinfection with a new E. coli strain, and 5% (2/40) had different E. coli strains at the two follow-up visits (one had reinfection followed by relapse, and the other had persistence followed by reinfection). In the placebo group the majority had E. coli at the first follow-up. PFGE showed that among these patients 96% (50/52) had persistence with the primary infecting E. coli strain and 4% (2/50) had different E. coli strains at the two follow-up visits (both had persistence followed by reinfection). The finding that the majority of UTIs at follow-up are caused by the primary infecting E. coli strain supports the theory of a vaginal and rectal reservoir but could also support the recent discovery that E. coli strains are able to persist in the bladder epithelium despite appropriate antibiotic treatment, constituting a reservoir for recurrent UTI.
INTRODUCTION
Urinary tract infection (UTI) is one of the most common bacterial infections of any organ system and accounts for significant morbidity and high medical costs (11, 32). The prevalence is influenced by gender and age—increasing with age and being most common among females except at early infancy (8). One study found that the incidence of symptomatic UTI among sexually active women aged 18 to 40 years was 0.5 to 0.7 per person-year (16). By the age of 24, one-third of the women were found to have had at least one physician-diagnosed UTI requiring antimicrobial treatment (11). The lifetime risk of UTI for women is reported to be in the range of 40 to 60% (11, 29). Recurrent UTI (RUTI) is common, i.e., 16 to 40% of women who experience an episode of UTI are reported to develop recurrence (10, 13, 19, 25). One-fourth of all women with a first UTI have been reported to develop a second UTI within 6 months (10), and within 12 months 50% of women have been reported to experience a recurrence (8, 19).
Considering the high figures for occurrence of RUTI, it is of considerable importance to understand the pathogenesis behind RUTI in order to be able to devise new treatment and preventive strategies in the future.
One aspect to consider regarding the pathogenesis of RUTI is whether it is attributable to reinfection with a new strain or relapse with the primary infecting strain.
Approximately 80 to 90% of all community-acquired UTIs and more than 30% of nosocomially acquired UTIs are caused by Escherichia coli (4, 9, 14, 17, 24). Previous studies dealing with RUTI and E. coli have concluded that RUTI is mainly attributable to reinfection with new strains (2, 5, 12, 19, 20, 25, 28, 31, 38, 45). All but three studies were based on serotyping or biochemical typing. The shortcomings of phenotypically based typing methods have led to the development of typing methods based on the microbial genotype or DNA sequence which give better typeability, reproducibility, and higher differentiation power. One smaller study with 23 women applied pulsed-field gel electrophoresis (PFGE) to E. coli strains from women with RUTI and questioned the traditional prevailing view of RUTI being caused by new strains and not the primary infecting strain (40). We present a study of 156 women with community-acquired lower UTIs caused by E. coli. The aim of the study was to determine by using PFGE whether UTI with E. coli at follow-up is due to a relapse or persistence of the primary infecting E. coli strain or to a reinfection with a new E. coli strain.
MATERIALS AND METHODS
Study population. The study is based on a strain collection from a multicenter, randomized, double-blind, placebo-controlled comparative study of different dosing regimens of pivmecillinam (piv-amdinocillin) for lower UTI. Eighteen Swedish primary health care centers prospectively included women with community-acquired lower UTI. Criteria for inclusion were women over 18 years of age with symptoms suggestive of lower UTI. Criteria for exclusion were pyelonephritis, antibiotic treatment for UTI within the last 30 days, genital infection, known anatomic defect in the urinary tract, diabetes, pregnancy, bladder catheter, incontinence requiring a catheter or pads, treatment with an unapproved drug within 3 months prior to inclusion, known/suspected penicillin allergy, or incapability of completing a patient diary or of medicating themselves (9). A total of 1,173 women, with a mean age of 43 years, were recruited. Eleven patients were excluded due to exclusion criteria, and 1,162 patients were included and randomized to pivmecillinam at 200 mg three times a day (TID) for 7 days, 200 mg twice a day (BID) for 7 days, 400 mg BID for 3 days, or placebo. Patients were evaluated clinically and bacteriologically at day 1 (inclusion), days 8 to 10 (first follow-up), and days 35 to 49 (second follow-up) (Table 1). The Ethics Committee of Umea University, Sweden, approved the study, and informed written consent to participate was obtained from the included patients.
As stated in the introduction the majority of UTIs are caused by E. coli, and in the parent study the majority (86%) of the cases with significant bacteriuria at follow-up among women with E. coli at inclusion were due to E. coli. Due to the predominance of E. coli this present study will focus on E. coli and is a subgroup of the parent study population focusing on patients with E. coli at inclusion and at follow-up. The subgroup (156 patients) was chosen according to the following criteria: from the pivmecillinam treatment group all patients showing significant E. coli bacteriuria at inclusion and at one or both of the follow-up visits were selected. From the placebo group a randomly selected group of 20 (if possible) patients from each category showing significant E. coli bacteriuria at inclusion and at one or both of the follow-up visits were selected (Table 2).
Specimen collection and processing. One cleanly voided urine sample was obtained for each patient on the day of inclusion and on the two scheduled follow-up visits. Urine was streaked onto a cystine-lactose-electrolyte-deficient agar plate (Acumedia Manufacturers, Inc., Baltimore, MD) with a 10-μl loop and incubated for 18 to 24 h at 35°C. Quantification was done as previously described (9). Identification was performed according to standard procedures as described previously followed by confirmation of identification of gram-negative rods by the API 20E (bioMerieux SA, Marcy l'Etoile, France) test (9). Antibiotic sensitivity testing was performed using multipoint inoculation on PDM-ASM agar (Biodisk, Solna, Sweden) against norfloxacin, amdinocillin, ampicillin, trimethoprim, nitrofurantoin, and cefadroxil. Breakpoints used were those recommended by the Swedish Reference Group for Antibiotics and its subcommittee on methodology (available at http://www.srga.org, accessed 20 August 2004). Subjects with isolates resistant to amdinocillin were excluded from analysis. Isolates were stored in freezing medium at –80°C for subsequent analyses.
Definitions. Significant bacteriuria was defined as a bacterial count of E. coli of 103 CFU/ml and 105 CFU/ml for symptomatic patients and asymptomatic patients, respectively. Mixed flora without one dominating species was classified as a negative culture, whereas in cases with mixed flora with one dominating species (bacterial count 10 times higher and 103 CFU/ml) this was defined as a uropathogen according to the above-mentioned criteria for E. coli.
Many terms regarding UTI have been applied with different definitions. The terms used in this study for UTI occurring at follow-up visits are applied according to the following definitions.
Bacterial persistence is used for UTI with the same strain as the primary infecting strain occurring after treatment of a UTI where there is no intervening negative culture separating the urine cultures.
Relapse is used for UTI with the same strain as the primary infecting strain occurring after treatment of a UTI when there is a negative urine culture or a culture with a strain different from the primary infecting strain separating the two urine cultures.
Reinfection is used for UTI with a new strain occurring after treatment of a UTI when there is a negative culture or no intervening culture separating the two urine cultures.
PFGE. The primary infecting E. coli strains from each woman were subjected to PFGE to analyze the study demographics. The paired E. coli isolates from the index episode and the follow-up visits were subjected to PFGE to analyze whether the UTIs at follow-up were caused by a new E. coli strain or by a strain the same as the primary infecting E. coli strain. Colonies from an overnight culture were suspended in 0.5 ml 0.9% saline to an optical density of 0.3 at a wavelength of 600 nm and resuspended in 0.5 ml TE buffer (10 mmol/liter Tris-HCl [pH 8.0], 100 mmol/liter EDTA). Chromosomal DNA was prepared in solid agarose plugs by mixing 0.5 ml bacterial cell suspension with an equal volume of 2% low-melting-point agarose (Sigma-Aldrich, Inc., St. Louis, Mo.) and then incubated overnight at 54°C in lysis buffer (50 mmol/liter Tris-HCl [pH 8.0], 50 mmol/liter EDTA, 1% laurylsarcosine, 1 mg of proteinase K/ml). The plugs were washed three times for 30 min in sterile distilled water and twice for 30 min in TE buffer (10 mmol/liter Tris-HCl [pH 8.0], 1 mmol/liter EDTA) at 50°C. A thin slice of plug was digested overnight with 50 U of XbaI restriction enzyme according to the manufacturer's instructions (New England Biolabs, Beverly, MA).
PFGE was performed with the CHEF DR-III (Bio-Rad Laboratories, Hercules, CA) system using a 1% SeaKem Gold Agarose gel (Cambrex Bio Science Rockland, Inc., Rockland, ME) in 2.5 liters 0.5x Tris-buffered EDTA running buffer (SSI Diagnostica, Hilleroed, Denmark). The electrophoretic conditions used were as follows: initial switch time, 2.0 s; final switch time, 35.0 s; run time, 21 h; gradient, 6 V/cm; angle, 120°; temperature, 14°C. A molecular size standard (Salmonella enterica serotype Braenderup H9812, CDC PFGE marker) (39) and a well-characterized internal control (Salmonella enterica serotype Senftenberg 99114159K) were processed at each run along with the E. coli isolates to be tested.
All PFGE profiles were subjected to data processing in BioNumerics 3.0 (Applied Maths, Sint-Martens-Latem, Belgium), using the unweighted-pair group method using arithmetic averages, the Dice coefficient, a position tolerance of 1%, and an optimization of 1%.
PFGE interpretation and definitions. Strains were defined as representing the same strain (being indistinguishable) if they possessed 100% similarity to the restriction fragment patterns of DNA (PFGE profile). Strains were defined as having a clonal relationship if they possessed 85% similarity to the PFGE profiles. The cutoff for clonal relationship was chosen because we considered isolates to have a clonal relationship if their PFGE profiles differed by changes consistent with a single genetic event, i.e., a point mutation or an insertion or deletion of DNA (44), which in this PFGE setup corresponded to approximately 85% similarity.
Statistics. The Fisher exact test and Cochran-Armitage trend test for comparison of proportions between the treatment groups were used. Analyses were performed by using SAS System for Windows V8 (SAS Institute Inc., Cary, NC).
RESULTS
A total of 156 patients with community-acquired lower UTI due to E. coli and significant bacteriuria with E. coli at one or both of the scheduled follow-up visits after treatment with pivmecillinam or placebo were studied. Fifty-four of the patients were from the placebo group, and 102 patients were from the pivmecillinam treatment groups distributed among the three different dosing regimens.
In total 80% of the patients in our study population had symptoms at follow-up. We chose to focus on the bacteriological aspects and did not stratify for the presence of symptoms in the analyses of our study.
Susceptibility testing showed that all but two of the primary infecting E. coli isolates were susceptible to amdinocillin. The subjects with isolates resistant to amdinocillin were excluded from PFGE analysis.
PFGE resolved approximately 15 to 20 distinct bands with fragments of 50 to 1,200 kb, which allowed a suitable discriminatory ability.
The PFGE profiles of the primary infecting E. coli strains were analyzed to describe the study demographics with respect to clonal relationship of the E. coli strains. Applying the criterion of 100% similarity of the PFGE profiles to the PFGE profiles of the primary infecting strain of the 156 patients revealed that 150 patients had an E. coli strain with a unique PFGE profile that was different among the patients. The remaining six patients were represented by three PFGE profiles, with two patients belonging to each of these. Applying the criterion of 85% similarity of the PFGE profiles to the PFGE profiles of the primary infecting strain of the 156 patients, it was found that 101 patients had an E. coli strain with a unique PFGE profile that was different for each patient. The remaining 55 patients were represented by 19 PFGE profiles with typically two to three patients, the maximum being five patients, belonging to each PFGE profile. It is not known whether the few persons sharing 85% similarity of the PFGE profile were epidemiologically related, but the UTI episodes of the patients were not temporally related. The PFGE results showed that E. coli strains from the study population were distinct without clonal dominance.
The criterion of 100% similarity was used when comparing the PFGE profile of the primary infecting E. coli strain with the PFGE profiles of E. coli strains from the follow-up visits for the same patients.
In the pivmecillinam treatment group 61 out of 102 patients presented with a negative urine culture at the first follow-up and E. coli at the second follow-up. PFGE showed that 77% (46/60) of the patients had UTI at follow-up due to relapse with the primary infecting E. coli strain and 23% (14/60) had UTI due to reinfection with a new E. coli strain.
The remaining 41 of the 102 patients in the pivmecillinam treatment group presented with E. coli at the first follow-up and either E. coli, a negative urine culture, or no culture (missing culture) at the second follow-up. PFGE typing revealed that 80% (32/40) of these patients had UTI at the follow-up visits due to persistence with the primary infecting E. coli strain and 15% (6/40) had UTI due to reinfection with a new E. coli strain. The remaining 5% (2/40) showed different E. coli strains at the two follow-up visits. PFGE typing showed that one patient had reinfection at the first follow-up and relapse at the second follow-up and the other patient had persistence at the first follow-up and reinfection at the second follow-up.
In the placebo group only two patients presented with a negative culture at the first follow-up and E. coli at the second follow-up. PFGE typing showed that one of the two patients had UTI at follow-up due to a relapse with the primary infecting E. coli strain and the other had UTI due to a reinfection with a new E. coli strain.
The remaining 52 out of 54 patients in the placebo group presented with E. coli at the first follow-up. PFGE typing illustrated that 96% (50/52) of these patients had UTI at follow-up due to persistence with the primary infecting E. coli strain. The remaining two patients (4%) showed different E. coli strains at the two follow-up visits. PFGE typing showed that they both had persistence at the first follow-up and reinfection at the second follow-up.
The exact figures regarding the PFGE results for the pivmecillinam treatment group and the placebo group can be seen in Table 3. Figure 1 shows a PFGE gel with PFGE profiles from 10 patients; all but one patient showed an E. coli strain at follow-up visits that was the same as the primary infecting strain.
DISCUSSION
This present study is a PFGE typing study of the E. coli strains from 156 selected women with uncomplicated community-acquired UTI with E. coli at inclusion and E. coli bacteriuria at the first and/or second follow-up visit after pivmecillinam or placebo treatment. PFGE showed that UTI at follow-up was mainly caused by the primary infecting E. coli strain.
The high frequency of UTI at follow-up caused by a strain indistinguishable from the primary infecting strain in a patient could not be explained by the hypothesis that only a single or a few clones caused UTI in the community of the study population. At a criterion of 85% similarity for clonal relationship 101 out of the 156 primary infecting strains showed a unique PFGE profile and the remaining 55 primary infecting strains were represented by 19 PFGE profiles. No clones were thus observed to be dominating in the study population and thereby constituting a possible explanation for the findings.
The microbiological success rates for the parent study (Table 1) showed that pivmecillinam was an effective drug, since bacteriological cure was seen in 78 to 89% of the patients treated with the drug in the three treatment arms in contrast to an effect of 28% in the placebo group at the first follow-up. Treatment failure cannot be excluded as a possible explanation for the UTI at follow-up caused by the primary infecting E. coli strain. Relapse with the primary infecting E. coli strain in the pivmecillinam arms of 7 days could represent suppressed persistence and thus a treatment failure due to the design of the study with the first follow-up taking place at days 8 to 10. This could inflate the same-strain recurrence rate.
In our study the rate of recurrence at the second follow-up caused by relapse with the primary infecting strain was found to be 81% (17/21) for the 3-day BID regimen, 76% (16/21) for the 7-day BID regimen, and 72% (13/18) for the 7-day TID regimen (Table 3). The differences between these rates for relapse for the different dosing regimens of pivmecillinam were not significant (P > 0.05), and the observed linear trend for relapse among the different treatment groups that could support the notion of a treatment failure as related to the duration of treatment in combination with the size of the dose as a possible explanation for the high rate of primary infecting E. coli strains at follow-up was not significant (P > 0.05).
Several studies have been published concerning RUTIs, and it has been discussed for a long time whether these are caused by the primary infecting strain/index strain or by a new strain. Many of these studies are in opposition to our findings and conclude that RUTIs are caused primarily by a new strain (2, 3, 13, 19, 20, 25, 28, 31, 38, 45).
Quite a few studies have used serotyping. These studies are reporting rates of relapses with the index strain in the range of 13 to 30% (2, 28, 31, 38, 45).
A biochemical fingerprinting system has also been applied. Jacobson et al. (20) found that 39% of RUTIs were caused by a relapse with the index strain compared to 33% in a study by Brauner et al., who also found that 53% of RUTIs were caused by an E. coli strain previously encountered (5).
Newer typing methods have been applied to studies of RUTI, and some of these also show results conflicting with ours. Two studies have used ribotyping. Bingen et al. (3) found that 1 out of 12 cases of recurrence was caused by the index strain. Ikaheimo et al. (19) used ribotyping for the bacteria related to the fraction of recurrences (43%) where phenotyping such as serotyping did not reveal an unequivocal identification. They found that only 33% of recurrences during a follow-up time of 1 year were caused by the index strain, and in looking selectively at the fraction of recurrences occurring 1 to 5 months after the index episode the rate was found to be 39 to 46%. Dot blot hybridization patterns of virulence factors have also been applied as a typing method, and Foxman et al. (12) found that 33% of second UTIs after E. coli UTI were caused by the same E. coli strain. This was done with reference to a previous study (13) that found an agreement between PFGE profiles and dot blot hybridization patterns of virulence factors. We question such a generalization because this method is too unspecific and therefore not qualified to be a reliable typing method. One study used randomly amplified polymorphic DNA-PCR and found that only 25% of recurrences were due to the index strain (25).
Few studies have used PFGE as a typing method. Jantunen et al. (22) found that the proportion of recurrences that represented a strain previously encountered (2-band difference of the PFGE profiles) was 65%. Two other studies have applied PFGE. Foxman et al. (13) showed that 50% of all second UTIs were caused by the index E. coli strain, and when restricted to E. coli recurrences the number was 58%. Russo et al. (40) found that 68% of RUTIs were caused by a strain with the same PFGE profile as previously identified in that person.
The reason for the difference between our study and these previous studies is not unequivocal, but these described studies differ from ours with respect to study population, including number, demographic characteristics of the patients and UTI category (cystitis and pyelonephritis) (2, 3, 5, 19, 20, 22, 25, 28, 31, 38, 40, 45), urine sample collection (3, 22), criteria for significant bacteriuria (2, 3, 5, 19, 22, 25, 28, 31, 40), treatment regimens (2, 5, 19, 20, 22, 28), length of follow-up time (2, 5, 12, 13, 20, 22, 25, 28, 31, 40), typing methods (2, 3, 5, 12, 19, 20, 25, 28, 31, 38, 45), and calculation/presentation of typing results (2, 28). All these factors might contribute to the observed difference between our findings and previously reported findings on the same-strain recurrence rate.
The age, gender, and predisposing condition of the study population can affect mechanisms responsible for recurrence of the same strain. Treatment regimens of various drugs, lengths, and doses can have different impacts on the reservoirs for E. coli and thus the relative frequency of same-strain versus different-strain recurrence. The methods of urine sample collection have impact on the contamination risk, which could affect the rate of same-strain recurrence. A lower cutoff for significant bacteriuria could contribute to a higher same-strain recurrence rate, as could the inclusion of asymptomatic bacteriuria. A long follow-up time and a RUTI occurring late in a long follow-up period reduce the chances of finding the same strain. Typing methods affect the rate of recurrence of the same strain according to typeability, reproducibility, and differentiation power of the method. We would expect that a lower typeability and differentiation power could lead to finding higher rates of same-strain recurrence. This was not the case for previous studies using phenotypic tests such as serotyping, biochemical fingerprinting, and some DNA-based methods. Regarding serotyping, a possible explanation could be a lower reproducibility because O and H antigens may be variably expressed. With respect to biochemical fingerprinting and DNA-based methods we do not have an explanation. The typing studies may differ with respect to whether they have limited the study population/their calculation of the same-strain recurrence rate to the total recurrent UTI cases or to those with E. coli recurrence only. The former would diminish and the latter might inflate the rate for same-strain recurrence.
The high frequency of RUTIs being caused by a strain identical to the primary infecting strain raises the question of whence these E. coli strains originate. The current prevailing theory regarding a reservoir for E. coli strains causing UTI is that they originate from the gastrointestinal tract flora (15). Several studies have supported the hypothesis of a fecal-perineal-periurethral route as well as a fecal-vaginal-periurethral route of infection (21, 36, 37, 40, 42, 46). It has been found that uropathogenic E. coli bacteria may be decreased in number (6, 27, 33) or eradicated (41) from the fecal reservoir after antibiotic treatment and may therefore not constitute a stable reservoir. With respect to pivmecillinam it has been shown to decrease the number of E. coli bacteria in the intestinal microflora (43), but for the actual dose and duration of therapy used in our study we cannot provide any information. Vaginal and rectal cultures were not done in our study, preventing any conclusions regarding these reservoirs. The lack of information regarding vaginal and rectal reservoirs for E. coli is an important limitation of our study. On the other hand, this type of strain collection would have increased the workload of the study immensely, and its relevance was not anticipated prior to the study. This could be the focus of a new prospective study.
While many RUTIs may be attributable to the above-mentioned pathogenesis of RUTI, a new potential reservoir has been discovered recently. It has been found that E. coli can invade bladder epithelial cells, a process mediated by the adhesion of the type 1 pilus, FimH (30, 34). E. coli can persist intracellularly in the bladder epithelium, replicating and forming intracellular niches termed bacterial "factories" that can mature into biofilms, thereby constituting a very potent and stable reservoir for RUTI (1, 18, 23, 34, 35, 41). The intracellularly located E. coli has been observed in mouse models for UTI and not in humans, but bacteria have been cultured from bladder tissue of women with recurrent UTIs during periods with an absence of bacteriuria (7). It has been shown that appropriate antibiotic treatment including amdinocillin that eradicates E. coli in the urine leaves the bladder reservoir unaffected (18, 26, 41) and capable of reestablishing an infection (26, 41). As for vaginal and rectal reservoirs we cannot provide data on the bladder reservoir. A future study with biopsy samples from the bladder would be optimal.
In conclusion, the findings in this study that the majority of UTIs at follow-up are attributable to bacterial persistence or relapse with the primary infecting E. coli strain support the theory of a rectal or vaginal reservoir but may also support the recent theory of a bladder reservoir for uropathogenic E. coli. The clinical implications of our results are unknown, but future research must focus on aspects of the reservoir for E. coli, including virulence factors related to this and finally possible treatment strategies to eradicate the reservoir and thereby reduce the frequency of RUTI.
ACKNOWLEDGMENTS
This work was supported by the Danish Medical Research Council, grant number 22-02-0373 ct/mp. The study from which the strains for this study originate was supported by LEO Pharma, Denmark.
REFERENCES
Anderson, G. G., J. J. Palermo, J. D. Schilling, R. Roth, J. Heuser, and S. J. Hultgren. 2003. Intracellular bacterial biofilm-like pods in urinary tract infections. Science 301:105-107.
Bergstrom, T., K. Lincoln, F. Orskov, I. Orskov, and J. Winberg. 1967. Studies of urinary tract infections in infancy and childhood. 8. Reinfection vs. relapse in recurrent urinary tract infections. Evaluation by means of identification of infecting organisms. J. Pediatr. 71:13-20.
Bingen, E., M. Rangaraj, and C. Safran. 1994. Ribotyping differentiates relapse from reinfection in the treatment failures of Escherichia coli urinary tract infections in children. Diagn. Microbiol. Infect. Dis. 18:263-265.
Bouza, E., R. San Juan, P. Munoz, A. Voss, J. Kluytmans, et al. 2001. A European perspective on nosocomial urinary tract infections II. Report on incidence, clinical characteristics and outcome (ESGNI-004 study). Clin. Microbiol. Infect. 7:532-542.
Brauner, A., S. H. Jacobson, and I. Kuhn. 1992. Urinary Escherichia coli causing recurrent infections—a prospective follow-up of biochemical phenotypes. Clin. Nephrol. 38:318-323.
Edlund, C., and C. E. Nord. 2000. Effect on the human normal microflora of oral antibiotics for treatment of urinary tract infections. J. Antimicrob. Chemother. 46(Suppl. 1):41-48.
Elliott, T. S., L. Reed, R. C. Slack, and M. C. Bishop. 1985. Bacteriology and ultrastructure of the bladder in patients with urinary tract infections. J. Infect. 11:191-199.
Ferry, S., L. G. Burman, and B. Mattsson. 1987. Urinary tract infection in primary health care in northern Sweden. I. Epidemiology. Scand. J. Prim. Health Care 5:123-128.
Ferry, S. A., S. E. Holm, H. Stenlund, R. Lundholm, and T. J. Monsen. 2004. The natural course of uncomplicated lower urinary tract infection in women illustrated by a randomized placebo controlled study. Scand. J. Infect. Dis. 36:296-301.
Foxman, B. 1990. Recurring urinary tract infection: incidence and risk factors. Am. J. Public Health 80:331-333.
Foxman, B., R. Barlow, H. D'Arcy, B. Gillespie, and J. D. Sobel. 2000. Urinary tract infection: self-reported incidence and associated costs. Ann. Epidemiol. 10:509-515.
Foxman, B., B. Gillespie, J. Koopman, L. Zhang, K. Palin, P. Tallman, J. V. Marsh, S. Spear, J. D. Sobel, M. J. Marty, and C. F. Marrs. 2000. Risk factors for second urinary tract infection among college women. Am. J. Epidemiol. 151:1194-1205.
Foxman, B., L. Zhang, P. Tallman, K. Palin, C. Rode, C. Bloch, B. Gillespie, and C. F. Marrs. 1995. Virulence characteristics of Escherichia coli causing first urinary tract infection predict risk of second infection. J. Infect. Dis. 172:1536-1541.
Haley, R. W., D. H. Culver, J. W. White, W. M. Morgan, T. G. Emori, V. P. Munn, and T. M. Hooton. 1985. The efficacy of infection surveillance and control programs in preventing nosocomial infections in US hospitals. Am. J. Epidemiol. 121:182-205.
Hooton, T. M. 2001. Recurrent urinary tract infection in women. Int. J. Antimicrob. Agents 17:259-268.
Hooton, T. M., D. Scholes, J. P. Hughes, C. Winter, P. L. Roberts, A. E. Stapleton, A. Stergachis, and W. E. Stamm. 1996. A prospective study of risk factors for symptomatic urinary tract infection in young women. N. Engl. J. Med. 335:468-474.
Hooton, T. M., and W. E. Stamm. 1997. Diagnosis and treatment of uncomplicated urinary tract infection. Infect. Dis. Clin. N. Am. 11:551-581.
Hvidberg, H., C. Struve, K. A. Krogfelt, N. Christensen, S. N. Rasmussen, and N. Frimodt-Moller. 2000. Development of a long-term ascending urinary tract infection mouse model for antibiotic treatment studies. Antimicrob. Agents Chemother. 44:156-163.
Ikaheimo, R., A. Siitonen, T. Heiskanen, U. Karkkainen, P. Kuosmanen, P. Lipponen, and P. H. Makela. 1996. Recurrence of urinary tract infection in a primary care setting: analysis of a 1-year follow-up of 179 women. Clin. Infect. Dis. 22:91-99.
Jacobson, S. H., I. Kuhn, and A. Brauner. 1992. Biochemical fingerprinting of urinary Escherichia coli causing recurrent infections in women with pyelonephritic renal scarring. Scand. J. Urol. Nephrol. 26:373-377.
Jantunen, M. E., H. Saxen, S. Lukinmaa, M. Ala-Houhala, and A. Siitonen. 2001. Genomic identity of pyelonephritogenic Escherichia coli isolated from blood, urine and faeces of children with urosepsis. J. Med. Microbiol. 50:650-652.
Jantunen, M. E., H. Saxen, E. Salo, and A. Siitonen. 2002. Recurrent urinary tract infections in infancy: relapses or reinfections J. Infect. Dis. 185:375-379.
Justice, S. S., C. Hung, J. A. Theriot, D. A. Fletcher, G. G. Anderson, M. J. Footer, and S. J. Hultgren. 2004. Differentiation and developmental pathways of uropathogenic Escherichia coli in urinary tract pathogenesis. Proc. Natl. Acad. Sci. USA 101:1333-1338.
Kahlmeter, G. 2000. The ECO.SENS Project: a prospective, multinational, multicentre epidemiological survey of the prevalence and antimicrobial susceptibility of urinary tract pathogens—interim report. J. Antimicrob. Chemother. 46(Suppl. 1):15-22.
Karkkainen, U. M., R. Ikaheimo, M. L. Katila, and A. Siitonen. 2000. Recurrence of urinary tract infections in adult patients with community-acquired pyelonephritis caused by E. coli: a 1-year follow-up. Scand. J. Infect. Dis. 32:495-499.
Kerrn, M. B., C. Struve, J. Blom, N. Frimodt-Moller, and K. A. Krogfelt. 2005. Intracellular persistence of Escherichia coli in urinary bladders from mecillinam-treated mice. J. Antimicrob. Chemother. 55:383-386.
Knothe, H. 1976. The influence of pivmecillinam on the human gut flora. Arzneimittelforschung 26:427-431. (In German.)
Kunin, C. M. 1970. A ten-year study of bacteriuria in schoolgirls: final report of bacteriologic, urologic, and epidemiologic findings. J. Infect. Dis. 122:382-393.
Kunin, C. M. 1994. Urinary tract infections in females. Clin. Infect. Dis. 18:1-10.
Martinez, J. J., M. A. Mulvey, J. D. Schilling, J. S. Pinkner, and S. J. Hultgren. 2000. Type 1 pilus-mediated bacterial invasion of bladder epithelial cells. EMBO J. 19:2803-2812.
McGeachie, J. 1966. Recurrent infection of the urinary tract: reinfection or recrudescence. Br. Med. J. 1:952-954.
Mobley, H. L., and J. W. Warren. 1996. Urinary tract infections: molecular pathogenesis and clinical management. ASM Press, Washington, D.C.
Moller, J. K., and A. Stenderup. 1984. The influence of long-term treatment with mecillinam on fecal Escherichia coli. Scand. J. Infect. Dis. 16:223-224.
Mulvey, M. A., Y. S. Lopez-Boado, C. L. Wilson, R. Roth, W. C. Parks, J. Heuser, and S. J. Hultgren. 1998. Induction and evasion of host defenses by type 1-piliated uropathogenic Escherichia coli. Science 282:1494-1497.
Mulvey, M. A., J. D. Schilling, and S. J. Hultgren. 2001. Establishment of a persistent Escherichia coli reservoir during the acute phase of a bladder infection. Infect. Immun. 69:4572-4579.
Navas-Nacher, E. L., F. Dardick, M. F. Venegas, B. E. Anderson, A. J. Schaeffer, and J. L. Duncan. 2001. Relatedness of Escherichia coli colonizing women longitudinally. Mol. Urol. 5:31-36.
Pfau, A., and T. Sacks. 1981. The bacterial flora of the vaginal vestibule, urethra and vagina in premenopausal women with recurrent urinary tract infections. J. Urol. 126:630-634.
Pryles, C. V., and A. Glagovsky. 1965. Serological characterization of Escherichia coli. Study in acute and recurrent urinary tract infections in infants and children. Pediatrics 36:219-224.
Romero, L., L. Lopez, L. Martinez-Martinez, B. Guerra, J. R. Hernandez, and A. Pascual. 2004. Characterization of the first CTX-M-14-producing Salmonella enterica serotype Enteritidis isolate. J. Antimicrob. Chemother. 53:1113-1114.
Russo, T. A., A. Stapleton, S. Wenderoth, T. M. Hooton, and W. E. Stamm. 1995. Chromosomal restriction fragment length polymorphism analysis of Escherichia coli strains causing recurrent urinary tract infections in young women. J. Infect. Dis. 172:440-445.
Schilling, J. D., R. G. Lorenz, and S. J. Hultgren. 2002. Effect of trimethoprim-sulfamethoxazole on recurrent bacteriuria and bacterial persistence in mice infected with uropathogenic Escherichia coli. Infect. Immun. 70:7042-7049.
Stamey, T. A., and C. C. Sexton. 1975. The role of vaginal colonization with enterobacteriaceae in recurrent urinary infections. J. Urol. 113:214-217.
Sullivan, A., C. Edlund, B. Svenungsson, L. Emtestam, and C. E. Nord. 2001. Effect of perorally administered pivmecillinam on the normal oropharyngeal, intestinal and skin microflora. J. Chemother. 13:299-308.
Tenover, F. C., R. D. Arbeit, R. V. Goering, P. A. Mickelsen, B. E. Murray, 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.
Winberg, J., H. J. Andersen, T. Bergstrom, B. Jacobsson, H. Larson, and K. Lincoln. 1974. Epidemiology of symptomatic urinary tract infection in childhood. Acta Paediatr. Scand. Suppl. 252:1-20.
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.(Karen Ejrnaes, Dorthe San)
Department of Clinical Microbiology, Hvidovre, Denmark
Department of Clinical Microbiology, Umea, Sweden
Umea Primary Health Care Centre, Umea, Sweden
Department of Medical Microbiology and Immunology, University of Gothenburg, Gothenburg, Sweden
ABSTRACT
The primary infecting Escherichia coli strains from 156 women with community-acquired uncomplicated urinary tract infection (UTI) randomized to pivmecillinam or placebo and the E. coli strains causing UTI at two follow-up visits were typed using pulsed-field gel electrophoresis (PFGE). In the pivmecillinam treatment group PFGE showed that among patients having a negative urine culture at the first follow-up 77% (46/60) had a relapse with the primary infecting E. coli strain and 23% (14/60) had reinfection with a new E. coli strain at the second follow-up. Among patients having E. coli at the first follow-up PFGE showed that 80% (32/40) had persistence with the primary infecting E. coli strain, 15% (6/40) had reinfection with a new E. coli strain, and 5% (2/40) had different E. coli strains at the two follow-up visits (one had reinfection followed by relapse, and the other had persistence followed by reinfection). In the placebo group the majority had E. coli at the first follow-up. PFGE showed that among these patients 96% (50/52) had persistence with the primary infecting E. coli strain and 4% (2/50) had different E. coli strains at the two follow-up visits (both had persistence followed by reinfection). The finding that the majority of UTIs at follow-up are caused by the primary infecting E. coli strain supports the theory of a vaginal and rectal reservoir but could also support the recent discovery that E. coli strains are able to persist in the bladder epithelium despite appropriate antibiotic treatment, constituting a reservoir for recurrent UTI.
INTRODUCTION
Urinary tract infection (UTI) is one of the most common bacterial infections of any organ system and accounts for significant morbidity and high medical costs (11, 32). The prevalence is influenced by gender and age—increasing with age and being most common among females except at early infancy (8). One study found that the incidence of symptomatic UTI among sexually active women aged 18 to 40 years was 0.5 to 0.7 per person-year (16). By the age of 24, one-third of the women were found to have had at least one physician-diagnosed UTI requiring antimicrobial treatment (11). The lifetime risk of UTI for women is reported to be in the range of 40 to 60% (11, 29). Recurrent UTI (RUTI) is common, i.e., 16 to 40% of women who experience an episode of UTI are reported to develop recurrence (10, 13, 19, 25). One-fourth of all women with a first UTI have been reported to develop a second UTI within 6 months (10), and within 12 months 50% of women have been reported to experience a recurrence (8, 19).
Considering the high figures for occurrence of RUTI, it is of considerable importance to understand the pathogenesis behind RUTI in order to be able to devise new treatment and preventive strategies in the future.
One aspect to consider regarding the pathogenesis of RUTI is whether it is attributable to reinfection with a new strain or relapse with the primary infecting strain.
Approximately 80 to 90% of all community-acquired UTIs and more than 30% of nosocomially acquired UTIs are caused by Escherichia coli (4, 9, 14, 17, 24). Previous studies dealing with RUTI and E. coli have concluded that RUTI is mainly attributable to reinfection with new strains (2, 5, 12, 19, 20, 25, 28, 31, 38, 45). All but three studies were based on serotyping or biochemical typing. The shortcomings of phenotypically based typing methods have led to the development of typing methods based on the microbial genotype or DNA sequence which give better typeability, reproducibility, and higher differentiation power. One smaller study with 23 women applied pulsed-field gel electrophoresis (PFGE) to E. coli strains from women with RUTI and questioned the traditional prevailing view of RUTI being caused by new strains and not the primary infecting strain (40). We present a study of 156 women with community-acquired lower UTIs caused by E. coli. The aim of the study was to determine by using PFGE whether UTI with E. coli at follow-up is due to a relapse or persistence of the primary infecting E. coli strain or to a reinfection with a new E. coli strain.
MATERIALS AND METHODS
Study population. The study is based on a strain collection from a multicenter, randomized, double-blind, placebo-controlled comparative study of different dosing regimens of pivmecillinam (piv-amdinocillin) for lower UTI. Eighteen Swedish primary health care centers prospectively included women with community-acquired lower UTI. Criteria for inclusion were women over 18 years of age with symptoms suggestive of lower UTI. Criteria for exclusion were pyelonephritis, antibiotic treatment for UTI within the last 30 days, genital infection, known anatomic defect in the urinary tract, diabetes, pregnancy, bladder catheter, incontinence requiring a catheter or pads, treatment with an unapproved drug within 3 months prior to inclusion, known/suspected penicillin allergy, or incapability of completing a patient diary or of medicating themselves (9). A total of 1,173 women, with a mean age of 43 years, were recruited. Eleven patients were excluded due to exclusion criteria, and 1,162 patients were included and randomized to pivmecillinam at 200 mg three times a day (TID) for 7 days, 200 mg twice a day (BID) for 7 days, 400 mg BID for 3 days, or placebo. Patients were evaluated clinically and bacteriologically at day 1 (inclusion), days 8 to 10 (first follow-up), and days 35 to 49 (second follow-up) (Table 1). The Ethics Committee of Umea University, Sweden, approved the study, and informed written consent to participate was obtained from the included patients.
As stated in the introduction the majority of UTIs are caused by E. coli, and in the parent study the majority (86%) of the cases with significant bacteriuria at follow-up among women with E. coli at inclusion were due to E. coli. Due to the predominance of E. coli this present study will focus on E. coli and is a subgroup of the parent study population focusing on patients with E. coli at inclusion and at follow-up. The subgroup (156 patients) was chosen according to the following criteria: from the pivmecillinam treatment group all patients showing significant E. coli bacteriuria at inclusion and at one or both of the follow-up visits were selected. From the placebo group a randomly selected group of 20 (if possible) patients from each category showing significant E. coli bacteriuria at inclusion and at one or both of the follow-up visits were selected (Table 2).
Specimen collection and processing. One cleanly voided urine sample was obtained for each patient on the day of inclusion and on the two scheduled follow-up visits. Urine was streaked onto a cystine-lactose-electrolyte-deficient agar plate (Acumedia Manufacturers, Inc., Baltimore, MD) with a 10-μl loop and incubated for 18 to 24 h at 35°C. Quantification was done as previously described (9). Identification was performed according to standard procedures as described previously followed by confirmation of identification of gram-negative rods by the API 20E (bioMerieux SA, Marcy l'Etoile, France) test (9). Antibiotic sensitivity testing was performed using multipoint inoculation on PDM-ASM agar (Biodisk, Solna, Sweden) against norfloxacin, amdinocillin, ampicillin, trimethoprim, nitrofurantoin, and cefadroxil. Breakpoints used were those recommended by the Swedish Reference Group for Antibiotics and its subcommittee on methodology (available at http://www.srga.org, accessed 20 August 2004). Subjects with isolates resistant to amdinocillin were excluded from analysis. Isolates were stored in freezing medium at –80°C for subsequent analyses.
Definitions. Significant bacteriuria was defined as a bacterial count of E. coli of 103 CFU/ml and 105 CFU/ml for symptomatic patients and asymptomatic patients, respectively. Mixed flora without one dominating species was classified as a negative culture, whereas in cases with mixed flora with one dominating species (bacterial count 10 times higher and 103 CFU/ml) this was defined as a uropathogen according to the above-mentioned criteria for E. coli.
Many terms regarding UTI have been applied with different definitions. The terms used in this study for UTI occurring at follow-up visits are applied according to the following definitions.
Bacterial persistence is used for UTI with the same strain as the primary infecting strain occurring after treatment of a UTI where there is no intervening negative culture separating the urine cultures.
Relapse is used for UTI with the same strain as the primary infecting strain occurring after treatment of a UTI when there is a negative urine culture or a culture with a strain different from the primary infecting strain separating the two urine cultures.
Reinfection is used for UTI with a new strain occurring after treatment of a UTI when there is a negative culture or no intervening culture separating the two urine cultures.
PFGE. The primary infecting E. coli strains from each woman were subjected to PFGE to analyze the study demographics. The paired E. coli isolates from the index episode and the follow-up visits were subjected to PFGE to analyze whether the UTIs at follow-up were caused by a new E. coli strain or by a strain the same as the primary infecting E. coli strain. Colonies from an overnight culture were suspended in 0.5 ml 0.9% saline to an optical density of 0.3 at a wavelength of 600 nm and resuspended in 0.5 ml TE buffer (10 mmol/liter Tris-HCl [pH 8.0], 100 mmol/liter EDTA). Chromosomal DNA was prepared in solid agarose plugs by mixing 0.5 ml bacterial cell suspension with an equal volume of 2% low-melting-point agarose (Sigma-Aldrich, Inc., St. Louis, Mo.) and then incubated overnight at 54°C in lysis buffer (50 mmol/liter Tris-HCl [pH 8.0], 50 mmol/liter EDTA, 1% laurylsarcosine, 1 mg of proteinase K/ml). The plugs were washed three times for 30 min in sterile distilled water and twice for 30 min in TE buffer (10 mmol/liter Tris-HCl [pH 8.0], 1 mmol/liter EDTA) at 50°C. A thin slice of plug was digested overnight with 50 U of XbaI restriction enzyme according to the manufacturer's instructions (New England Biolabs, Beverly, MA).
PFGE was performed with the CHEF DR-III (Bio-Rad Laboratories, Hercules, CA) system using a 1% SeaKem Gold Agarose gel (Cambrex Bio Science Rockland, Inc., Rockland, ME) in 2.5 liters 0.5x Tris-buffered EDTA running buffer (SSI Diagnostica, Hilleroed, Denmark). The electrophoretic conditions used were as follows: initial switch time, 2.0 s; final switch time, 35.0 s; run time, 21 h; gradient, 6 V/cm; angle, 120°; temperature, 14°C. A molecular size standard (Salmonella enterica serotype Braenderup H9812, CDC PFGE marker) (39) and a well-characterized internal control (Salmonella enterica serotype Senftenberg 99114159K) were processed at each run along with the E. coli isolates to be tested.
All PFGE profiles were subjected to data processing in BioNumerics 3.0 (Applied Maths, Sint-Martens-Latem, Belgium), using the unweighted-pair group method using arithmetic averages, the Dice coefficient, a position tolerance of 1%, and an optimization of 1%.
PFGE interpretation and definitions. Strains were defined as representing the same strain (being indistinguishable) if they possessed 100% similarity to the restriction fragment patterns of DNA (PFGE profile). Strains were defined as having a clonal relationship if they possessed 85% similarity to the PFGE profiles. The cutoff for clonal relationship was chosen because we considered isolates to have a clonal relationship if their PFGE profiles differed by changes consistent with a single genetic event, i.e., a point mutation or an insertion or deletion of DNA (44), which in this PFGE setup corresponded to approximately 85% similarity.
Statistics. The Fisher exact test and Cochran-Armitage trend test for comparison of proportions between the treatment groups were used. Analyses were performed by using SAS System for Windows V8 (SAS Institute Inc., Cary, NC).
RESULTS
A total of 156 patients with community-acquired lower UTI due to E. coli and significant bacteriuria with E. coli at one or both of the scheduled follow-up visits after treatment with pivmecillinam or placebo were studied. Fifty-four of the patients were from the placebo group, and 102 patients were from the pivmecillinam treatment groups distributed among the three different dosing regimens.
In total 80% of the patients in our study population had symptoms at follow-up. We chose to focus on the bacteriological aspects and did not stratify for the presence of symptoms in the analyses of our study.
Susceptibility testing showed that all but two of the primary infecting E. coli isolates were susceptible to amdinocillin. The subjects with isolates resistant to amdinocillin were excluded from PFGE analysis.
PFGE resolved approximately 15 to 20 distinct bands with fragments of 50 to 1,200 kb, which allowed a suitable discriminatory ability.
The PFGE profiles of the primary infecting E. coli strains were analyzed to describe the study demographics with respect to clonal relationship of the E. coli strains. Applying the criterion of 100% similarity of the PFGE profiles to the PFGE profiles of the primary infecting strain of the 156 patients revealed that 150 patients had an E. coli strain with a unique PFGE profile that was different among the patients. The remaining six patients were represented by three PFGE profiles, with two patients belonging to each of these. Applying the criterion of 85% similarity of the PFGE profiles to the PFGE profiles of the primary infecting strain of the 156 patients, it was found that 101 patients had an E. coli strain with a unique PFGE profile that was different for each patient. The remaining 55 patients were represented by 19 PFGE profiles with typically two to three patients, the maximum being five patients, belonging to each PFGE profile. It is not known whether the few persons sharing 85% similarity of the PFGE profile were epidemiologically related, but the UTI episodes of the patients were not temporally related. The PFGE results showed that E. coli strains from the study population were distinct without clonal dominance.
The criterion of 100% similarity was used when comparing the PFGE profile of the primary infecting E. coli strain with the PFGE profiles of E. coli strains from the follow-up visits for the same patients.
In the pivmecillinam treatment group 61 out of 102 patients presented with a negative urine culture at the first follow-up and E. coli at the second follow-up. PFGE showed that 77% (46/60) of the patients had UTI at follow-up due to relapse with the primary infecting E. coli strain and 23% (14/60) had UTI due to reinfection with a new E. coli strain.
The remaining 41 of the 102 patients in the pivmecillinam treatment group presented with E. coli at the first follow-up and either E. coli, a negative urine culture, or no culture (missing culture) at the second follow-up. PFGE typing revealed that 80% (32/40) of these patients had UTI at the follow-up visits due to persistence with the primary infecting E. coli strain and 15% (6/40) had UTI due to reinfection with a new E. coli strain. The remaining 5% (2/40) showed different E. coli strains at the two follow-up visits. PFGE typing showed that one patient had reinfection at the first follow-up and relapse at the second follow-up and the other patient had persistence at the first follow-up and reinfection at the second follow-up.
In the placebo group only two patients presented with a negative culture at the first follow-up and E. coli at the second follow-up. PFGE typing showed that one of the two patients had UTI at follow-up due to a relapse with the primary infecting E. coli strain and the other had UTI due to a reinfection with a new E. coli strain.
The remaining 52 out of 54 patients in the placebo group presented with E. coli at the first follow-up. PFGE typing illustrated that 96% (50/52) of these patients had UTI at follow-up due to persistence with the primary infecting E. coli strain. The remaining two patients (4%) showed different E. coli strains at the two follow-up visits. PFGE typing showed that they both had persistence at the first follow-up and reinfection at the second follow-up.
The exact figures regarding the PFGE results for the pivmecillinam treatment group and the placebo group can be seen in Table 3. Figure 1 shows a PFGE gel with PFGE profiles from 10 patients; all but one patient showed an E. coli strain at follow-up visits that was the same as the primary infecting strain.
DISCUSSION
This present study is a PFGE typing study of the E. coli strains from 156 selected women with uncomplicated community-acquired UTI with E. coli at inclusion and E. coli bacteriuria at the first and/or second follow-up visit after pivmecillinam or placebo treatment. PFGE showed that UTI at follow-up was mainly caused by the primary infecting E. coli strain.
The high frequency of UTI at follow-up caused by a strain indistinguishable from the primary infecting strain in a patient could not be explained by the hypothesis that only a single or a few clones caused UTI in the community of the study population. At a criterion of 85% similarity for clonal relationship 101 out of the 156 primary infecting strains showed a unique PFGE profile and the remaining 55 primary infecting strains were represented by 19 PFGE profiles. No clones were thus observed to be dominating in the study population and thereby constituting a possible explanation for the findings.
The microbiological success rates for the parent study (Table 1) showed that pivmecillinam was an effective drug, since bacteriological cure was seen in 78 to 89% of the patients treated with the drug in the three treatment arms in contrast to an effect of 28% in the placebo group at the first follow-up. Treatment failure cannot be excluded as a possible explanation for the UTI at follow-up caused by the primary infecting E. coli strain. Relapse with the primary infecting E. coli strain in the pivmecillinam arms of 7 days could represent suppressed persistence and thus a treatment failure due to the design of the study with the first follow-up taking place at days 8 to 10. This could inflate the same-strain recurrence rate.
In our study the rate of recurrence at the second follow-up caused by relapse with the primary infecting strain was found to be 81% (17/21) for the 3-day BID regimen, 76% (16/21) for the 7-day BID regimen, and 72% (13/18) for the 7-day TID regimen (Table 3). The differences between these rates for relapse for the different dosing regimens of pivmecillinam were not significant (P > 0.05), and the observed linear trend for relapse among the different treatment groups that could support the notion of a treatment failure as related to the duration of treatment in combination with the size of the dose as a possible explanation for the high rate of primary infecting E. coli strains at follow-up was not significant (P > 0.05).
Several studies have been published concerning RUTIs, and it has been discussed for a long time whether these are caused by the primary infecting strain/index strain or by a new strain. Many of these studies are in opposition to our findings and conclude that RUTIs are caused primarily by a new strain (2, 3, 13, 19, 20, 25, 28, 31, 38, 45).
Quite a few studies have used serotyping. These studies are reporting rates of relapses with the index strain in the range of 13 to 30% (2, 28, 31, 38, 45).
A biochemical fingerprinting system has also been applied. Jacobson et al. (20) found that 39% of RUTIs were caused by a relapse with the index strain compared to 33% in a study by Brauner et al., who also found that 53% of RUTIs were caused by an E. coli strain previously encountered (5).
Newer typing methods have been applied to studies of RUTI, and some of these also show results conflicting with ours. Two studies have used ribotyping. Bingen et al. (3) found that 1 out of 12 cases of recurrence was caused by the index strain. Ikaheimo et al. (19) used ribotyping for the bacteria related to the fraction of recurrences (43%) where phenotyping such as serotyping did not reveal an unequivocal identification. They found that only 33% of recurrences during a follow-up time of 1 year were caused by the index strain, and in looking selectively at the fraction of recurrences occurring 1 to 5 months after the index episode the rate was found to be 39 to 46%. Dot blot hybridization patterns of virulence factors have also been applied as a typing method, and Foxman et al. (12) found that 33% of second UTIs after E. coli UTI were caused by the same E. coli strain. This was done with reference to a previous study (13) that found an agreement between PFGE profiles and dot blot hybridization patterns of virulence factors. We question such a generalization because this method is too unspecific and therefore not qualified to be a reliable typing method. One study used randomly amplified polymorphic DNA-PCR and found that only 25% of recurrences were due to the index strain (25).
Few studies have used PFGE as a typing method. Jantunen et al. (22) found that the proportion of recurrences that represented a strain previously encountered (2-band difference of the PFGE profiles) was 65%. Two other studies have applied PFGE. Foxman et al. (13) showed that 50% of all second UTIs were caused by the index E. coli strain, and when restricted to E. coli recurrences the number was 58%. Russo et al. (40) found that 68% of RUTIs were caused by a strain with the same PFGE profile as previously identified in that person.
The reason for the difference between our study and these previous studies is not unequivocal, but these described studies differ from ours with respect to study population, including number, demographic characteristics of the patients and UTI category (cystitis and pyelonephritis) (2, 3, 5, 19, 20, 22, 25, 28, 31, 38, 40, 45), urine sample collection (3, 22), criteria for significant bacteriuria (2, 3, 5, 19, 22, 25, 28, 31, 40), treatment regimens (2, 5, 19, 20, 22, 28), length of follow-up time (2, 5, 12, 13, 20, 22, 25, 28, 31, 40), typing methods (2, 3, 5, 12, 19, 20, 25, 28, 31, 38, 45), and calculation/presentation of typing results (2, 28). All these factors might contribute to the observed difference between our findings and previously reported findings on the same-strain recurrence rate.
The age, gender, and predisposing condition of the study population can affect mechanisms responsible for recurrence of the same strain. Treatment regimens of various drugs, lengths, and doses can have different impacts on the reservoirs for E. coli and thus the relative frequency of same-strain versus different-strain recurrence. The methods of urine sample collection have impact on the contamination risk, which could affect the rate of same-strain recurrence. A lower cutoff for significant bacteriuria could contribute to a higher same-strain recurrence rate, as could the inclusion of asymptomatic bacteriuria. A long follow-up time and a RUTI occurring late in a long follow-up period reduce the chances of finding the same strain. Typing methods affect the rate of recurrence of the same strain according to typeability, reproducibility, and differentiation power of the method. We would expect that a lower typeability and differentiation power could lead to finding higher rates of same-strain recurrence. This was not the case for previous studies using phenotypic tests such as serotyping, biochemical fingerprinting, and some DNA-based methods. Regarding serotyping, a possible explanation could be a lower reproducibility because O and H antigens may be variably expressed. With respect to biochemical fingerprinting and DNA-based methods we do not have an explanation. The typing studies may differ with respect to whether they have limited the study population/their calculation of the same-strain recurrence rate to the total recurrent UTI cases or to those with E. coli recurrence only. The former would diminish and the latter might inflate the rate for same-strain recurrence.
The high frequency of RUTIs being caused by a strain identical to the primary infecting strain raises the question of whence these E. coli strains originate. The current prevailing theory regarding a reservoir for E. coli strains causing UTI is that they originate from the gastrointestinal tract flora (15). Several studies have supported the hypothesis of a fecal-perineal-periurethral route as well as a fecal-vaginal-periurethral route of infection (21, 36, 37, 40, 42, 46). It has been found that uropathogenic E. coli bacteria may be decreased in number (6, 27, 33) or eradicated (41) from the fecal reservoir after antibiotic treatment and may therefore not constitute a stable reservoir. With respect to pivmecillinam it has been shown to decrease the number of E. coli bacteria in the intestinal microflora (43), but for the actual dose and duration of therapy used in our study we cannot provide any information. Vaginal and rectal cultures were not done in our study, preventing any conclusions regarding these reservoirs. The lack of information regarding vaginal and rectal reservoirs for E. coli is an important limitation of our study. On the other hand, this type of strain collection would have increased the workload of the study immensely, and its relevance was not anticipated prior to the study. This could be the focus of a new prospective study.
While many RUTIs may be attributable to the above-mentioned pathogenesis of RUTI, a new potential reservoir has been discovered recently. It has been found that E. coli can invade bladder epithelial cells, a process mediated by the adhesion of the type 1 pilus, FimH (30, 34). E. coli can persist intracellularly in the bladder epithelium, replicating and forming intracellular niches termed bacterial "factories" that can mature into biofilms, thereby constituting a very potent and stable reservoir for RUTI (1, 18, 23, 34, 35, 41). The intracellularly located E. coli has been observed in mouse models for UTI and not in humans, but bacteria have been cultured from bladder tissue of women with recurrent UTIs during periods with an absence of bacteriuria (7). It has been shown that appropriate antibiotic treatment including amdinocillin that eradicates E. coli in the urine leaves the bladder reservoir unaffected (18, 26, 41) and capable of reestablishing an infection (26, 41). As for vaginal and rectal reservoirs we cannot provide data on the bladder reservoir. A future study with biopsy samples from the bladder would be optimal.
In conclusion, the findings in this study that the majority of UTIs at follow-up are attributable to bacterial persistence or relapse with the primary infecting E. coli strain support the theory of a rectal or vaginal reservoir but may also support the recent theory of a bladder reservoir for uropathogenic E. coli. The clinical implications of our results are unknown, but future research must focus on aspects of the reservoir for E. coli, including virulence factors related to this and finally possible treatment strategies to eradicate the reservoir and thereby reduce the frequency of RUTI.
ACKNOWLEDGMENTS
This work was supported by the Danish Medical Research Council, grant number 22-02-0373 ct/mp. The study from which the strains for this study originate was supported by LEO Pharma, Denmark.
REFERENCES
Anderson, G. G., J. J. Palermo, J. D. Schilling, R. Roth, J. Heuser, and S. J. Hultgren. 2003. Intracellular bacterial biofilm-like pods in urinary tract infections. Science 301:105-107.
Bergstrom, T., K. Lincoln, F. Orskov, I. Orskov, and J. Winberg. 1967. Studies of urinary tract infections in infancy and childhood. 8. Reinfection vs. relapse in recurrent urinary tract infections. Evaluation by means of identification of infecting organisms. J. Pediatr. 71:13-20.
Bingen, E., M. Rangaraj, and C. Safran. 1994. Ribotyping differentiates relapse from reinfection in the treatment failures of Escherichia coli urinary tract infections in children. Diagn. Microbiol. Infect. Dis. 18:263-265.
Bouza, E., R. San Juan, P. Munoz, A. Voss, J. Kluytmans, et al. 2001. A European perspective on nosocomial urinary tract infections II. Report on incidence, clinical characteristics and outcome (ESGNI-004 study). Clin. Microbiol. Infect. 7:532-542.
Brauner, A., S. H. Jacobson, and I. Kuhn. 1992. Urinary Escherichia coli causing recurrent infections—a prospective follow-up of biochemical phenotypes. Clin. Nephrol. 38:318-323.
Edlund, C., and C. E. Nord. 2000. Effect on the human normal microflora of oral antibiotics for treatment of urinary tract infections. J. Antimicrob. Chemother. 46(Suppl. 1):41-48.
Elliott, T. S., L. Reed, R. C. Slack, and M. C. Bishop. 1985. Bacteriology and ultrastructure of the bladder in patients with urinary tract infections. J. Infect. 11:191-199.
Ferry, S., L. G. Burman, and B. Mattsson. 1987. Urinary tract infection in primary health care in northern Sweden. I. Epidemiology. Scand. J. Prim. Health Care 5:123-128.
Ferry, S. A., S. E. Holm, H. Stenlund, R. Lundholm, and T. J. Monsen. 2004. The natural course of uncomplicated lower urinary tract infection in women illustrated by a randomized placebo controlled study. Scand. J. Infect. Dis. 36:296-301.
Foxman, B. 1990. Recurring urinary tract infection: incidence and risk factors. Am. J. Public Health 80:331-333.
Foxman, B., R. Barlow, H. D'Arcy, B. Gillespie, and J. D. Sobel. 2000. Urinary tract infection: self-reported incidence and associated costs. Ann. Epidemiol. 10:509-515.
Foxman, B., B. Gillespie, J. Koopman, L. Zhang, K. Palin, P. Tallman, J. V. Marsh, S. Spear, J. D. Sobel, M. J. Marty, and C. F. Marrs. 2000. Risk factors for second urinary tract infection among college women. Am. J. Epidemiol. 151:1194-1205.
Foxman, B., L. Zhang, P. Tallman, K. Palin, C. Rode, C. Bloch, B. Gillespie, and C. F. Marrs. 1995. Virulence characteristics of Escherichia coli causing first urinary tract infection predict risk of second infection. J. Infect. Dis. 172:1536-1541.
Haley, R. W., D. H. Culver, J. W. White, W. M. Morgan, T. G. Emori, V. P. Munn, and T. M. Hooton. 1985. The efficacy of infection surveillance and control programs in preventing nosocomial infections in US hospitals. Am. J. Epidemiol. 121:182-205.
Hooton, T. M. 2001. Recurrent urinary tract infection in women. Int. J. Antimicrob. Agents 17:259-268.
Hooton, T. M., D. Scholes, J. P. Hughes, C. Winter, P. L. Roberts, A. E. Stapleton, A. Stergachis, and W. E. Stamm. 1996. A prospective study of risk factors for symptomatic urinary tract infection in young women. N. Engl. J. Med. 335:468-474.
Hooton, T. M., and W. E. Stamm. 1997. Diagnosis and treatment of uncomplicated urinary tract infection. Infect. Dis. Clin. N. Am. 11:551-581.
Hvidberg, H., C. Struve, K. A. Krogfelt, N. Christensen, S. N. Rasmussen, and N. Frimodt-Moller. 2000. Development of a long-term ascending urinary tract infection mouse model for antibiotic treatment studies. Antimicrob. Agents Chemother. 44:156-163.
Ikaheimo, R., A. Siitonen, T. Heiskanen, U. Karkkainen, P. Kuosmanen, P. Lipponen, and P. H. Makela. 1996. Recurrence of urinary tract infection in a primary care setting: analysis of a 1-year follow-up of 179 women. Clin. Infect. Dis. 22:91-99.
Jacobson, S. H., I. Kuhn, and A. Brauner. 1992. Biochemical fingerprinting of urinary Escherichia coli causing recurrent infections in women with pyelonephritic renal scarring. Scand. J. Urol. Nephrol. 26:373-377.
Jantunen, M. E., H. Saxen, S. Lukinmaa, M. Ala-Houhala, and A. Siitonen. 2001. Genomic identity of pyelonephritogenic Escherichia coli isolated from blood, urine and faeces of children with urosepsis. J. Med. Microbiol. 50:650-652.
Jantunen, M. E., H. Saxen, E. Salo, and A. Siitonen. 2002. Recurrent urinary tract infections in infancy: relapses or reinfections J. Infect. Dis. 185:375-379.
Justice, S. S., C. Hung, J. A. Theriot, D. A. Fletcher, G. G. Anderson, M. J. Footer, and S. J. Hultgren. 2004. Differentiation and developmental pathways of uropathogenic Escherichia coli in urinary tract pathogenesis. Proc. Natl. Acad. Sci. USA 101:1333-1338.
Kahlmeter, G. 2000. The ECO.SENS Project: a prospective, multinational, multicentre epidemiological survey of the prevalence and antimicrobial susceptibility of urinary tract pathogens—interim report. J. Antimicrob. Chemother. 46(Suppl. 1):15-22.
Karkkainen, U. M., R. Ikaheimo, M. L. Katila, and A. Siitonen. 2000. Recurrence of urinary tract infections in adult patients with community-acquired pyelonephritis caused by E. coli: a 1-year follow-up. Scand. J. Infect. Dis. 32:495-499.
Kerrn, M. B., C. Struve, J. Blom, N. Frimodt-Moller, and K. A. Krogfelt. 2005. Intracellular persistence of Escherichia coli in urinary bladders from mecillinam-treated mice. J. Antimicrob. Chemother. 55:383-386.
Knothe, H. 1976. The influence of pivmecillinam on the human gut flora. Arzneimittelforschung 26:427-431. (In German.)
Kunin, C. M. 1970. A ten-year study of bacteriuria in schoolgirls: final report of bacteriologic, urologic, and epidemiologic findings. J. Infect. Dis. 122:382-393.
Kunin, C. M. 1994. Urinary tract infections in females. Clin. Infect. Dis. 18:1-10.
Martinez, J. J., M. A. Mulvey, J. D. Schilling, J. S. Pinkner, and S. J. Hultgren. 2000. Type 1 pilus-mediated bacterial invasion of bladder epithelial cells. EMBO J. 19:2803-2812.
McGeachie, J. 1966. Recurrent infection of the urinary tract: reinfection or recrudescence. Br. Med. J. 1:952-954.
Mobley, H. L., and J. W. Warren. 1996. Urinary tract infections: molecular pathogenesis and clinical management. ASM Press, Washington, D.C.
Moller, J. K., and A. Stenderup. 1984. The influence of long-term treatment with mecillinam on fecal Escherichia coli. Scand. J. Infect. Dis. 16:223-224.
Mulvey, M. A., Y. S. Lopez-Boado, C. L. Wilson, R. Roth, W. C. Parks, J. Heuser, and S. J. Hultgren. 1998. Induction and evasion of host defenses by type 1-piliated uropathogenic Escherichia coli. Science 282:1494-1497.
Mulvey, M. A., J. D. Schilling, and S. J. Hultgren. 2001. Establishment of a persistent Escherichia coli reservoir during the acute phase of a bladder infection. Infect. Immun. 69:4572-4579.
Navas-Nacher, E. L., F. Dardick, M. F. Venegas, B. E. Anderson, A. J. Schaeffer, and J. L. Duncan. 2001. Relatedness of Escherichia coli colonizing women longitudinally. Mol. Urol. 5:31-36.
Pfau, A., and T. Sacks. 1981. The bacterial flora of the vaginal vestibule, urethra and vagina in premenopausal women with recurrent urinary tract infections. J. Urol. 126:630-634.
Pryles, C. V., and A. Glagovsky. 1965. Serological characterization of Escherichia coli. Study in acute and recurrent urinary tract infections in infants and children. Pediatrics 36:219-224.
Romero, L., L. Lopez, L. Martinez-Martinez, B. Guerra, J. R. Hernandez, and A. Pascual. 2004. Characterization of the first CTX-M-14-producing Salmonella enterica serotype Enteritidis isolate. J. Antimicrob. Chemother. 53:1113-1114.
Russo, T. A., A. Stapleton, S. Wenderoth, T. M. Hooton, and W. E. Stamm. 1995. Chromosomal restriction fragment length polymorphism analysis of Escherichia coli strains causing recurrent urinary tract infections in young women. J. Infect. Dis. 172:440-445.
Schilling, J. D., R. G. Lorenz, and S. J. Hultgren. 2002. Effect of trimethoprim-sulfamethoxazole on recurrent bacteriuria and bacterial persistence in mice infected with uropathogenic Escherichia coli. Infect. Immun. 70:7042-7049.
Stamey, T. A., and C. C. Sexton. 1975. The role of vaginal colonization with enterobacteriaceae in recurrent urinary infections. J. Urol. 113:214-217.
Sullivan, A., C. Edlund, B. Svenungsson, L. Emtestam, and C. E. Nord. 2001. Effect of perorally administered pivmecillinam on the normal oropharyngeal, intestinal and skin microflora. J. Chemother. 13:299-308.
Tenover, F. C., R. D. Arbeit, R. V. Goering, P. A. Mickelsen, B. E. Murray, 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.
Winberg, J., H. J. Andersen, T. Bergstrom, B. Jacobsson, H. Larson, and K. Lincoln. 1974. Epidemiology of symptomatic urinary tract infection in childhood. Acta Paediatr. Scand. Suppl. 252:1-20.
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.(Karen Ejrnaes, Dorthe San)