INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Urinary tract infection (UTI) is a frequent medical complication during the initial medical and rehabilitation period after spinal-cord injury (SCI) and traumatic brain injury (TBI) and continues to be a problem throughout the life of many spinal-cord-injured individuals. The urine of patients with neuropathic bladders frequently contains bacteria, and Escherichia coli is among the most frequent bladder colonizers (1, 5, 6). These bacteria may produce symptoms of UTI or may produce only asymptomatic bacteriuria (ABU). Bacteria associated with ABU are often left untreated and may even be beneficial in preventing symptomatic infection by more virulent organisms (16). Little is known at present about the virulence properties of E. coli bacteria that cause symptomatic UTI in patients with neuropathic bladders or how these strains may differ from benign colonizing strains. In the present study, approved by the Institutional Review Board of the hospital at which it was performed, we investigated selected virulence properties of E. coli isolates from SCI and TBI patients with UTI and ABU. Our hypothesis was that E. coli isolates from patients with symptomatic UTI possess virulence properties that may be used to distinguish them from other E. coli strains.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Patient characteristics. Subjects were selected from patients admitted to a rehabilitation hospital between January 1994 and December 1996 with a diagnosis of SCI or TBI who were treated with intermittent catheterization and acquired at least one colonization of the urinary tract with E. coli. Twenty-five patients developed 28 symptomatic UTIs due to E. coli during that time. Of these 16 males and 9 females, 4 had TBI and 21 had SCI (12 tetraplegics and 9 paraplegics); the mean age was 39 (range, 7 to 75). Twenty-seven patients were colonized 29 times with E. coli but remained asymptomatic. Of these 15 males and 12 females, 1 had TBI and 19 had SCI (9 tetraplegics and 10 paraplegics); the mean age was 42 (range, 14 to 70). Symptomatic UTI was defined as a fever of at least 38.5°C (101°F) plus one or more of the following symptoms in a patient with a urine colony count of >100,000 CFU per ml: voiding between catheterizations which had not previously occurred, otherwise unexplained increase in spasticity, pyuria (>10 leukocytes per high-power field), or hematuria. Such UTIs were treated with an antibiotic chosen from the antibiogram report indicating sensitivities to antibiotics. All patients had documented sterile urine after discontinuation of the antibiotics.

Collection of specimens. All patients while on intermittent catheterization had a catheterized urine specimen collected under sterile conditions once weekly. Isolates from patients with symptomatic UTI used in this study were collected prior to initiation of antimicrobial therapy. These specimens were kept on ice and transported to the clinical microbiology laboratory within 20 min. Samples were streaked for single colonies on MacConkey agar, sheep blood agar, and Columbia CNA agar plates (BBL, Cockeysville, Md.). E. coli isolates were identified by standard microbiological methods. All E. coli isolates identified from colonizations where E. coli was the only species, or the predominant species (>99%), present were subjected to further analysis as described below. In five instances, isolates from the same patient were collected from separate infection or colonization events. Such isolates were verified as unique based on differences in antibiogram biotype and/or pulsed-field gel electrophoretic analysis of endonuclease-digested whole-cell DNA. The charts of the patients from whose specimens the E. coli isolates grew were also reviewed based on the criteria listed above to determine whether or not each patient had a UTI and was in fact treated. If the patient's manifestations met the criteria and the patient was treated, the colonization was counted as symptomatic.

Determination of phenotypes. Methods used for assay of hemolysin expression, serum resistance, and D-mannose-resistant hemagglutination (MRHA) of human erythrocytes have been described previously and were used without modification (9, 12). Strains were identified as prototrophic or auxotrophic by testing for capacity to grow on minimal salts agar supplemented with 0.4% glucose (4).

Colony blot hybridization. Colony DNA blots were prepared from a single colony of each isolate by the steam-alkaline lysis method of Maas (14). Hybridization probes were purified as restriction endonuclease fragments from recombinant plasmids or synthesized as PCR amplification products from plasmid-containing strains. Specific hybridization probes used for detection of dra, pap, hly, and pil genes have been described elsewhere (2, 9, 17). Probes specific for focH and sfaS were prepared by PCR amplification of recombinant-DNA strains HB101(pPIL110-54) and HB101(pANN801-13), respectively, kindly provided by J. Hacker. The sequences of PCR primers used were 5' GACGTGGATACGACGATTACTG 3' and 5' TACGCATAGGTATAGGTGAC 3'. The probe for ucaA was prepared by PCR amplification of Proteus mirabilis HU1069 (3). Primer sequences were 5' CTCATAAGCGATGGTGTAATGAACTGTAGC 3' and 5' TATGACGGTACAATTACTTTTACTGGAAA 3'. Probes were radiolabeled with a random priming kit (Pharmacia). Hybridizations were conducted at high stringency (in 1× SSC [0.15 M NaCl plus 0.015 M sodium citrate] at 68°C).

Antimicrobial resistance. The disk diffusion method was used to determine antimicrobial susceptibility. Antimicrobials tested included amikacin, ampicillin, ampicillin-sublactam, augmentin, cefazolin, ceftazidime, ceftriaxone, cefuroxime, cephalothin, ciprofloxacin, gentamicin, nitrofurantoin, norfloxacin, piperacillin, tetracycline, tobramycin and trimethoprim-sulfamethoxazole.

Statistical methods. Statistical comparisons were made by chi-square analysis.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Single urinary tract (UT) E. coli isolates from patients experiencing symptomatic UTI or ABU were examined for expression of selected virulence phenotypes and for the presence of virulence-associated genes. The results are summarized in Tables 1 and 2, respectively.

Hemagglutination. Isolates were tested for MRHA, the ability to agglutinate human erythrocytes in the presence of the receptor analog D-mannose. The MRHA⁺ phenotype was more frequent among symptomatic-UTI isolates than among ABU isolates. The frequencies of P pilus genes among symptomatic-UTI isolates and ABU strains were similar. Two E. coli isolates from patients with ABU possessed pap genes but did not express MRHA. One ABU isolate was MRHA⁺ and possessed dra, but not pap, genes. Symptomatic-UTI and ABU isolates were also tested for the presence of other UT-associated adherence genes, including genes for type 1 (pil), S (sfa), Dr (dra), type 1C (foc), and Uca (uca) pili. For all except pil, the probes detected homologous sequences at low frequency. One or more of the adherence genes were present in 48% of symptomatic-UTI isolates and 36% of ABU isolates. As reported in numerous other studies, type 1 pilus genes (pil) were found in nearly all isolates; due to the nearly ubiquitous nature of type 1 pili and the pil⁺ genotype among the Enterobacteriaceae, no further analysis was pursued (2).

Hemolysin. The Hly⁺ phenotype was significantly more prevalent among E. coli isolates from patients with symptomatic UTI than among ABU strains. Isolates were also probed for hemolysin-specific DNA sequences. The frequency of hly genes among symptomatic-UTI strains was greater than that among ABU isolates, but the difference was not significant. Three Hly hly⁺ isolates and one that was Hly⁺ but lacked hly were observed among E. coli isolates from patients with ABU.

Serum resistance. No significant difference was observed in frequency of resistance to human serum between symptomatic-UTI and ABU E. coli isolates. Values were similar to the reported frequency of 52% for normal fecal strains.

Antimicrobial resistance. No significant difference was observed in frequency of resistance to one or more antimicrobial agents between E. coli isolates from patients with symptomatic UTI and ABU strains. The number of antimicrobial agents (of 17 tested) to which these 49 isolates were resistant ranged from 1 to 9, with a mean of 4.3.

Auxotrophy. No significant difference was observed in frequency of requirement for nutritional supplements for growth between symptomatic-UTI and ABU isolates. Values are similar to reported values for frequency of auxotrophy among normal fecal E. coli isolates (35%) and isolates from non-SCI subjects with symptomatic UTI (58%).

Associations. The combined strain set (symptomatic UTI plus ABU), which represents E. coli isolates that colonize the neuropathic bladder, was analyzed for associations between virulence factors. This analysis revealed that pap⁺ isolates were more frequently hly⁺ and expressed the hemolytic phenotype (hly was present in 13 of 16 pap⁺ isolates but only in 1 of 35 isolates lacking pap) (P < 0.005). A trend toward increased antimicrobial resistance among pap⁺ strains was also noted (8 of 16 pap⁺ strains versus 8 of 33 strains lacking pap were resistant) (P < 0.1). In addition, a strong association was observed between the auxotrophic phenotype and sensitivity to antimicrobial agents (16 of 18 auxotrophic isolates versus 15 of 29 prototrophic isolates were sensitive) (P < 0.01).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Hemolysin is a cytolytic enzyme secreted by many E. coli isolates from patients with extraintestinal infections. Our results suggest that hemolysin may contribute to clinical symptoms of UTI in SCI patients. This conclusion is consistent with results of an earlier study of UT colonization in young women without SCI wherein the hemolytic phenotype was found at significantly increased frequency among symptomatic-UTI versus ABU isolates (8).

Bacterial adherence is widely held to contribute to colonization of the normal UT. Our results suggest that D-mannose-resistant adherence and the pap⁺ genotype also were associated with colonization of the UT in our patient group. The present results are in contrast with a previous study of catheter-associated E. coli bacteriuria in geriatric subjects wherein no correlation was found between MRHA or pap genotype and UT colonization (15). However, these studies may not be directly comparable because subjects in the previous study had indwelling urinary catheters in place at least 30 days prior to the study, while the subjects in the present study were managed with intermittent catheterization. The frequency of the MRHA⁺ phenotype was also significantly greater among isolates from patients with symptomatic infection in our study group compared with the reported frequency of MRHA⁺ E. coli in bowel flora of the general population (16%) (P < 0.001) (7, 10). One interpretation may be that D-mannose-resistant adherence contributes to E. coli urovirulence in this patient group. Alternately, the increased presence of MRHA⁺ bacteria among symptomatic-UTI isolates may reflect their possibly increased presence in the bowel flora of our study group. However, numerous studies have shown a low frequency of MRHA among fecal isolates from a variety of subject populations. Additional studies comparing the frequency of virulence factors among fecal isolates of SCI patients with those for other subject groups will be required to address this possibility.

Consistent with numerous published reports showing that E. coli type 1 pili contribute to bladder colonization, pil genes were detected in a majority of both symptomatic-UTI and ABU E. coli isolates in this study. Other UT-associated adherence genes representing S, type 1C, and Dr pili were found at reduced frequency in all SCI isolates, as has been reported for E. coli isolates from the UTs of non-SCI patients (10). E. coli strains were tested for the presence of uca genes, previously described only in P. mirabilis, because of reported similarities between the Uca structural protein and a protein associated with the E. coli G hemagglutinin (3). uca genes were found at low frequency among urine isolates from SCI and TBI subjects. Additional studies will be required to determine whether uca genes in E. coli promote uroepithelial cell adherence as they do in P. mirabilis.

The frequencies of other tested properties, including serum resistance, antimicrobial resistance, and auxotrophy, among symptomatic-UTI and ABU E. coli isolates were not significantly different. However, associations were found between these properties and other virulence-associated traits. We observed, as have others, a close correlation between pap and hly, most likely the result of these traits being linked genetically on the E. coli chromosome (9, 11). We also found a significant association between auxotrophy and sensitivity to antimicrobial agents commonly used for treatment of UTI. This finding suggests possible selection of auxotrophy among sensitive clinical isolates. The mechanism for selection is currently unclear. Our results did not confirm an increased frequency of auxotrophic bacteria among UT isolates compared with normal fecal E. coli isolates as reported in studies of non-SCI subjects (13).

In conclusion, we have determined that E. coli isolated from patients with symptomatic UTI undergoing intermittent catheterization may be distinguished from ABU isolates and normal fecal bacteria on the basis of their virulence characteristics. Our results point to the need for further research to determine if virulence factors such as hemolysin can guide the clinician in deciding whether or not to treat for E. coli colonization by indicating which isolates pose a higher risk of causing a symptomatic UTI under certain conditions. The results also suggest that future vaccines directed against selected bacterial virulence antigens may be of value in reducing the frequency of symptomatic UTI in this high-risk patient group.