Bacterial Interference for Prevention of Urinary Tract Infection: A Prospective, Randomized, Placebo-Controlled, Double-Blind Pilot Trial
Rabih O. Darouiche
John I. Thornby
William H. Donovan
Richard A. Hull
Baylor College of Medicine
Institute for Rehabilitation and Research
Received 17 May 2005; accepted 6 July 2005;
electronically published 13 October 2005. Baylor College of Medicine
, 1333 Moursund Ave., Ste. A221,
Houston, TX 77030
Michael E. Debakey Veterans Affairs Medical Center
(See the editorial commentary by Ronald on pages 1535-6) This prospective, randomized, placebo-controlled, doubleblind pilot trial examined the efficacy of bacterial interference in preventing urinary tract infection (UTI) in 27 patients with spinal cord injury. Patients whose bladders became colonized with Escherichia coli 83972 were half as likely (P p .01) than noncolonized patients to develop UTI during the subsequent year.
Approximately 247,000 persons currently live with
complications of spinal cord injury (SCI) in the United States .
Urinary tract infection (UTI), the most common infection in this
population, is associated with major medical and economic
sequelae and is difficult to prevent . Moreover, systemic
antibiotic prophylaxis [5, 6], local antimicrobial-utilizing
approaches (including daily care of the urethral meatus) [7, 8],
disinfection of the collection bag [9, 10], and irrigation of the
catheter and bladder [11, 12] have not been demonstrated to
adequately prevent catheter-related UTI in either patients with
SCI or able-bodied persons . The limited success of such
traditional approaches prompted our interest in exploring the
potential preventive efficacy of bacterial interference.
Earlier investigations have indicated that inoculation of the
neurogenic bladder of patients with SCI with a nonpathogenic
strain (i.e., although it has the genes for 4 fimbriae types [pap,
fim, foc, and uca] and hemolysin [hly], it does not express any)
of Escherichia coli 83972 can result in long-term asymptomatic
bladder colonization . Although a small, prospective
clinical trial has shown that patients with SCI had significantly
lower rates of symptomatic UTI while their bladders were
colonized with E. coli 83972, compared with their baseline prestudy
rates , that initial study was limited by its nonrandomized,
single-arm, open-label design. The objective of this pilot clinical
trial was to examine in a more scientifically robust fashion the
efficacy and safety of this approach with bacterial interference.
Patients and methods. This prospective, randomized,
double-blind, placebo-controlled trial was approved by the
institutional review board. Patients were randomly assigned in a 3:
1 ratio to have their bladders inoculated with either E. coli
83972 (experimental group) or sterile normal saline (control
Adult patients with SCI for a duration of 1 year who had
a neurogenic bladder that required indwelling (transurethral or
suprapubic) or intermittent catheter drainage and experienced
frequent, symptomatic episodes of UTI ( 2 episodes during
the preceding year) were eligible for enrollment. Exclusion
criteria included supravesical urinary diversion, vesicoureteral
reflux, obstructing urolithiasis, an indwelling nephrostomy
catheter, extraurogenital infections that require prolonged antibiotic
therapy, uncontrolled diabetes mellitus, and
immunosuppression. After obtaining informed consent, a medical history was
obtained, and physical examination, urine culture, radiologic
evaluation with plain radiography of the abdomen, renal
ultrasound, and urodynamic study were performed to check for
urolithiasis and vesicoureteral reflux.
Because previous experience indicated that successful
colonization with E. coli 83972 was more likely achieved by
inoculating sterile rather than nonsterile bladders [13, 16], patients
with positive baseline urine culture results completed a 1-week
course of properly selected antibiotics 23 days before
inoculation. After inserting a new catheter and emptying the bladder,
a 30-mL aliquot of either E. coli 83972 suspension (106 cfu/mL
of saline) or sterile saline was instilled into the bladder and
allowed to dwell for 2 h. Each cycle of inoculations consisted
of bladder instillations twice per day for 3 consecutive days.
Patients were successfully kept unaware of the nature of
solutions instilled into their bladders.
Urine samples for culture were obtained by means of freshly
inserted catheters at 1 week after bladder inoculation, then
monthly thereafter for 1 year. Successful bladder colonization
was defined by the growth ( 103 cfu/mL) of E. coli 83972 in
Age, mean years (range)
Level of spinal cord injury
Type of bladder catheter
Mean no. of UTIs in the past
Table 1. Clinical characteristics of patients inoculated with
Escherichia coli 83972 or sterile normal saline.
NOTE. Data are no. of patients, unless otherwise indicated. NS, not
significant; UTI, urinary tract infection.
urine cultures for 1 month. To distinguish E. coli 83972 from
other E. coli strains, we relied on both antibiotic susceptibility
patterns and restriction fragment-length polymorphism
analysis that used PFGE. Patients who became initially colonized
with E. coli 83972 and then cleared this organism from their
bladders did not undergo further inoculations.
The primary outcome of UTI was diagnosed by a blinded
evaluator (R.O.D.) for patients who were unaware of their
randomization group by the presence of significant bacteriuria
( 105cfu/mL) and pyuria (110 WBCs/high power field) plus
1 of the following signs and symptoms, for which no other
etiology could be identified: fever (oral temperature, 137.8 C),
sprapubic or flank discomfort, bladder spasm, increased
spasticity, and worsening dysreflexia [17, 18].
The clinical characteristics of patients were compared by
using Fishers exact test or the x2 test for categorical variables
and Students t test for continuous variables. Fishers exact test
was used to compare the proportions of patients in different
groups who developed 1 episode of UTI. The numbers of
episodes of UTI were compared using the Wilcoxon rank sum
test. The proportions of patients who were free of UTI as a
function of the time elapsed between bladder inoculation and
the first episode of UTI were compared between patients who
became colonized with E. coli 83972 and noncolonized patients,
using the log-rank test on the Kaplan-Meier estimates. All P
values were based on 2-tailed tests of significance.
Results. A total of 27 male patients were randomized to
the experimental group (21 patients) or placebo group (6
patients). The 2 groups of patients had similar clinical
characteristics (table 1). Thirteen (62%) of 21 patients in the
experimental group became colonized with E. coli 83972 for 1
month, including 4 patients who remained colonized
throughout the 12-month monitoring period and 9 others who lost
the inoculated organism after an average of 3.5 months.
Repeated urine cultures performed during periods of colonization
revealed 3 growth patterns: E. coli 83972 that always grew alone
(3 patients); E. coli 83972 that initially grew alone, then in
combination with other organisms (5 patients); and E. coli
83972 that always grew in combination with other organisms
(5 patients). Regardless, E. coli 83972 that usually grew at a
concentration higher than or similar to concentrations of
Overall, 13 (62%) of 21 patients in the experimental group
and all 6 patients in the control group developed at least 1
episode of UTI during the 1-year follow-up period (P p .07).
The number of UTI episodes that occurred during the
followup period was significantly lower (P p .036) in the
experimental group (mean, 1.6 episodes) than in the control group
(mean, 3.5 episodes).
Because there were no differences in the clinical
characteristics and the incidence of UTI among the 8 patients in the
experimental group who could not be colonized with E. coli
83972 and the 6 patients in the control group, all 14 patients
were considered to be noncolonized for statistical comparisons.
Patients colonized with E. coli 83972 were significantly less likely
than noncolonized patients to develop at least 1 episode of UTI
during the 1-year follow-up period (6 [46%] of 13 patients vs.
13 [93%] of 14 patients; P p .01). Figure 1 shows the
KaplanMeier estimates of the risk of UTI, by time elapsed since bladder
inoculation in colonized versus noncolonized patients and
illustrates that bladder colonization with E. coli 83972 was
protective against infection (P p .002, by the log-rank test).
A single patient developed autonomic dysreflexia 30 min
after instillation of placebo solution and recovered completely
after the catheter was unclamped and the inoculum was drained
from the bladder. None of the 27 patients had clinical evidence
of septicemia following bladder inoculations. Only 1 of 13
colonized patients developed UTI while E. coli 93972 still existed
in the bladder; however, because that episode of UTI was
characterized by polymicrobial growth of other organisms
(including Pseudomonas aeruginosa that had previously caused UTI)
in urine cultures, the patient was diagnosed and treated for
UTI due to P. aeruginosa.
Discussion. The results of this controlled pilot study
provide the first valid proof that bacterial interference based on
deliberate bladder colonization with E. coli 83972 protects
against UTI in patients with SCI. The observed 2-fold reduction
(46% vs. 93%) in the likelihood of UTI developing in colonized
versus noncolonized patients surpasses the magnitude of
protection afforded by all other potentially preventive approaches,
including antimicrobial-coated catheters, whose activity can be
overwhelmed by the high concentration of urinary pathogens
. Because only 1 of 13 colonized patients developed UTI
while E. coli 93972 still existed in the bladder, we have embarked
on genetic alteration of E. coli 83972 in an effort to prolong
the duration of bladder colonization (1 patient remained
colonized for almost 3 years) and, we hope, to further augment
protection against infection. Although the exact mechanism(s)
by which E. coli 83972 exerts its protective impact is still unclear,
possibilities include secretion of substances that can kill or
inhibit growth of other pathogens, physical blockage of binding
sites on the uroepithelium, and alteration of the nutritional or
pH level in the surrounding environment.
The lack of evidence in this study of septicemia and UTI
attributed to E. coli 83972 after bladder inoculations, coupled
with our previous finding of no adverse impact of chronic E.
coli 83972 colonization on renal function, support the safety
of this preventive approach . Potential limitations of this
small-sized study include the inclusion of adult men only, the
nonblinding of the nurse who performed bladder inoculations,
and the inability to perform cost-benefit analysis. Regardless,
these promising results have allowed us to initiate a large
National Institutes of Healthsponsored, prospective, randomized,
placebo-controlled, double-blind, multicenter clinical trial to
confirm the efficacy and safety and assess the cost-benefit of
bacterial interference in more diverse SCI populations.
Financial support. Rehabilitation Research and Development Service
grant from the Department of Veterans Affairs (B241OR) to R.O.D., and
the Paralyzed Veterans of America Spinal Cord Research Foundation
(RFA302) and the US Public Health Service (HD35856) to R.A.H.
Potential conflicts of interest. All authors: no conflicts.