375

Recurrent Urinary Tract Infections in Infancy: Relapses or Reinfections?

M. E. Jantunen,1,2 H. Saxén,1 E. Salo,1 and A. Siitonen2 1
Hospital for Children and Adolescents, University of Helsinki,
and 2Laboratory of Enteric Pathogens, National Public Health
Institute, Helsinki, Finland

Seventeen infants with an index episode of pyelonephritis caused by Escherichia coli were
monitored for 18 months for recurrent urinary tract infections (UTIs). All the infants had at
least 1 recurrent UTI caused by the same pathogen. Twenty-six recurrent UTI episodes were re-
corded. The 40 E. coli strains available were analyzed by multiplex polymerase chain reaction
for 3 alleles (classes I– III) of the papG gene and by pulsed-field gel electrophoresis (PFGE)
after genomial digestion by Xba I. Of the 17 index strains, 12 (71%) carried the papG gene;
67% of these strains had class II alleles. In recurrent UTI isolates, the papG-positive E. coli ap-
peared in 16 (70%) of 23 isolates. The proportion of all recurrent isolates available that rep-
resented a strain previously encountered (indistinguishable or highly similar in PFGE) in the
same infant was 65%. Our results suggest that most recurrent UTIs in infants are endogenous
relapses rather than reinfections caused by new organisms.

Symptomatic urinary tract infection (UTI) recurrences are Methods

common among children [1], with young children and those
with urinary tract abnormalities especially at risk [2, 3]. Recur- Study population. In a prospective multicenter follow-up
rent UTIs may cause long-term consequences for children, be- study, 180 infants (age 1– 24 months) were monitored for recurrent
cause the number of pyelonephritogenic attacks increases the UTI for 18 months after the index episode of acute PN. The charac-
risk of renal scarring [1, 4]. teristics of the study population have been presented elsewhere [9].
In brief, there were 107 girls and 73 boys. Infants were recruited
The P-fimbrial phenotype of Escherichia coli strains causing
from 5 hospitals in southern Finland between April 1995 and Feb-
pyelonephritis (PN) was found to be associated with recurrent
ruary 1998. For the present study, only the infants fulfilling all the
UTI episodes [5]. P fimbriae are known to promote bacterial bind- following criteria were included: the index PN was caused by E. coli;
ing onto uroepithelial cells via the adhesins encoded by 3 alleles at least 1 recurrent UTI was caused by E. coli; the recurrent urinary
(classes I– III) of the papG gene [6]. The class III papG gene is strain was available for analysis in addition to the index PN strain;
commonly found in E. coli strains that cause cystitis [7] and the and a full 18-month follow-up was completed. Of the 180 infants,
class II papG gene in strains that cause PN [8]. 17 were included in the analysis.
Although several studies of adults have compared the charac- PN was defined as the combination of a positive urine culture
teristics of urinary isolates that cause recurrent UTIs after an (growth >105 bacteria/mL of identical species in 2 voided sterile-
index episode of PN or cystitis, no genotypic analysis of recurrent bag specimens or any growth in a suprapubic bladder aspirate) and
UTI isolates from children with PN has been published. Because the presence of at least 2 of the following 3 criteria: C-reactive pro-
tein .25 mg/L, temperature >38.0
C, and pyuria (leukocyte count
the papG genotype alone is a weak predictor of strain identity, we
.10 cells/mm3). Patients who did not fulfil the criteria for PN but
also applied pulsed-field gel electrophoresis (PFGE) to study the
who had symptomatic, microbiologically confirmed UTI were clas-
identity of isolates from infants with recurrent UTIs after PN. sified as having cystitis. A recurrent UTI was defined as a UTI epi-
sode after successful treatment of PN or cystitis (a negative urine
culture, absence of pyuria, and absence of fever). Asymptomatic
bacteriuria was not considered a recurrent UTI.
Received 19 March 2001; revised 30 September 2001; electronically pub- Radiological imaging studies. All infants underwent a renal
lished 9 January 2002. ultrasound. In addition, either a voiding cystourethrography or a
Informed consent was obtained from the parent(s) or guardian(s) of radioisotope voiding cystourethrography was performed on each
participants at entry into the study. The study protocol was approved by the
institutional review board and ethical committee of each participating unit. infant to define possible vesicoureteral reflux (VUR). Additional
Financial support: The Foundation for Pediatric Research, Helsinki, imaging studies were performed whenever needed [9].
Finland; the Research Funds of Helsinki University Central Hospital The infants were classified into 2 groups according to the clinical
(grant TYH 8237); and the Maud Kuistila Foundation, Finland. significance of the urinary tract findings. One group consisted of in-
Reprints or correspondence: Dr. Anja Siitonen, National Public Health
Institute, Laboratory of Enteric Pathogens, Mannerheimintie 166, FIN-00300
fants with normal anatomy or with insignificant urinary tract ab-
Helsinki, Finland ([email protected]). normalities, such as VUR grade I or II or bladder dysfunction.
The other group consisted of infants with significant urinary tract
The Journal of Infectious Diseases 2002;185:375–9
q 2002 by the Infectious Diseases Society of America. All rights reserved. abnormalities that caused increased susceptibility to infections,
0022-1899/2002/18503-0013$02.00 such as VUR grade III or IV or obstructive anomaly.

Downloaded from https://academic.oup.com/jid/article-abstract/185/3/375/895950

by guest
on 07 June 2018
 376 Jantunen et al. JID 2002;185 (1 February)

Treatment and follow-up. The infants received intravenous Results

antimicrobials according to the study protocol. The recommen-
dations were to treat children either with cefuroxime or, for infants There were 26 recurrent episodes of UTI detected among 17
,6 weeks old, with ampicillin in combination with aminoglyco- infants, of whom 8 were boys (median age at entry, 2.6 months;
side, until the infants were afebrile. Cefuroxime alone was used range, 1.5– 3.1 months) and 9 were girls (median age at entry,
for 16 of 17 infants. One infant first received cefuroxime and later 9.0 months; range, 1.0– 19.0 months). Five infants had 2 recur-
ceftriaxone. Therapy was tailored according to the results of the rent infections, and 2 infants had 3 recurrent infections. Of the re-
antimicrobial susceptibility tests as soon as they were available. currences, 6 were PN and 20 were cystitis. All the PN strains and
The parenteral therapy was continued with an oral regimen, to 17 of 20 cystitis strains were available for analysis. Thus, 23 E.
complete a 10-day course of treatment. All infants received prophy- coli strains were available for genotypic analysis (40 strains, in-
lactic antimicrobials (in most cases trimethoprim) until imaging
cluding 17 strains from index PN; table 1).
studies were performed. This initial prophylaxis usually lasted 4–6
The shortest interval between 2 recurrent infections was 24
weeks. Infants with significant urinary tract abnormalities con-
tinued on prophylactic antimicrobials to prevent recurrent UTIs, days. The first recurrence occurred at a median of 3.0 months
as did infants with normal anatomy who developed recurrent (range, 0.8–6.1 months) after the initial infection. The majority
infections. (22/26 [85%]) of the recurrences appeared within 6 months
All infants were monitored for 18 months. General practitioners at after the index episode of PN.
health centers monitored mainly infants who did not show any struc- Of the 17 index strains, 12 (71%) carried the papG gene; of
tural or functional abnormalities in imaging studies. However, 1 visit those, 67% were positive for class II alleles. In recurrent UTI iso-
to a hospital 12 months after the index episode of PN was rec- lates, the papG-positive E. coli appeared in 16 (70%) of 23 iso-
ommended. Infants with urinary tract abnormalities were followed lates; of those, 63% were positive for class II alleles. The first
up at 6, 12, and 18 months in their respective hospitals after the recurrent isolate available was distinguishable from the index
index episode of PN. All the infants were scheduled for regular isolate (at least a 3-band difference in the PFGE profile) in on-
urine testing at 2, 3, 4, 5, 6, 12, and 18 months at a health care center
ly 6 (35%) of 17 infants. In 7 (41%) infants, the isolate was indis-
or hospital during the follow-up visits. In addition, parents were en-
tinguishable from the initial pyelonephritogenic isolate (no
couraged to consult their general practitioners whenever a UTI was
suspected. All recurrent UTIs were microbiologically confirmed. difference in PFGE profile) (figure 1). An isolate highly similar
Standard clinical practice was followed in the treatment of recurrent (1 or 2-fragment difference in the PFGE profile) to the index iso-
infections. All infections, including UTIs and respiratory infections, late was found in 4 (24%) of 17 infants. Four of 6 PN strains and 7
antimicrobial prescriptions, and visits to health care facilities were of 11 cystitis strains were indistinguishable from or highly simi-
recorded by parents for 18 months. In addition, the parents were con- lar to the index strain.
tacted by telephone (by M.E.J.) every 3– 6 months, to increase the The proportion of all recurrent isolates available that rep-
motivation of the families to remain in the study. resented a strain previously encountered (indistinguishable or
Urine samples. Urine samples were collected by use of bag highly similar in PFGE) in the same infant was 15 (65%) of 23.
specimens or suprapubic bladder aspirates before initiation of These isolates included 2 of the “multiple recurrence” isolates
therapy. Subsequently, the samples were cultured, and the pathogens
(nos. 15 and 16), which, although unrelated to the index strain,
were identified and their sensitivity to several antimicrobials were
represented repeat isolates of the preceding recurrence isolate.
determined. The treatment was later modified, if needed, according
to susceptibility tests. All positive cultures were sent to the National
Of those infants (n ¼ 7) who had .1 recurrence, there were 13
Public Health Institute, Helsinki, where identification of the patho- of 16 isolates available. Of these isolates, 8 (62%) of 13 rep-
gen was verified by standard methods, and they were stored in skim resented a strain previously encountered in the same infant.
milk at 270
C for further analyses [10]. Urinary tract anatomy and function were normal or nearly nor-
Polymerase chain reaction (PCR) assay and PFGE. E. coli iso- mal in 10 (83%) of 12 infants when the recurrent UTI was caused
lates were first analyzed by multiplex PCR for class I, II, and III by indistinguishable or highly similar isolates (nos. 1, 2, 4–9, 11,
alleles of the papG gene [11]. Subsequently, each isolate was ana- and 12) (table 1) and in 4 (80%) of 5 infants when the causative
lyzed by PFGE after Xba I restriction of chromosomal DNA to com- agent was unrelated to the index isolate (nos. 14– 17). Seven
pare their macrorestriction profiles [12]. The isolates that were (58%) of 12 infants who had recurrent infections caused by indis-
identical to the index PN isolate were termed “indistinguishable.” tinguishable or highly similar isolates did not receive prophylac-
The isolates with a 1- or 2-fragment difference, compared with the
tic antimicrobials to prevent recurrent UTIs. On the other hand, 1
infant’s index PN isolate, were termed “highly similar.” The isolates
of 5 infants who had a recurrent UTI caused by an unrelated isolate
with a .2 fragment difference, compared with the index isolate,
were termed “unrelated.”
did not receive prophylactic antimicrobials. To analyze whether
Statistical methods. Data are expressed as mean ^ SD or me- the use of frequent antibiotic courses was associated with occur-
dian (range). Analysis of variance and analysis of covariance were rence of unrelated strains in recurrent UTIs, we also collected data
used for multiple comparisons. P , :05 was considered to be signifi- on the overall use of other antimicrobials. The median number of
cant. Analyses were performed by use of the SPSS software package antimicrobials, mainly to treat respiratory infections, was 3.0
(version 7.5 for Windows; SPSS). (range, 0–14) in addition to prophylactic antimicrobial treatment

Downloaded from https://academic.oup.com/jid/article-abstract/185/3/375/895950

by guest
on 07 June 2018
 JID 2002;185 (1 February) Recurrent Urinary Tract Infections 377

Table 1. Genetic clones identified by polymerase chain reaction Discussion

( papG classes I, II, and III) and by pulsed-field gel electrophoresis
(PFGE) of Escherichia coli isolates from 17 infants with an index epi- Forty E. coli isolates from index PN or recurrent UTIs were at
sode of pyelonephritis (PN) and recurrent urinary tract infection. first investigated for P fimbriae encoding papG genes and sub-
Interval, papG sequently characterized by PFGE to study the strain identity.
Infant months allele PFGE profilea Infectionb The strains were considered either indistinguishable (identical
1 0 III profile in PFGE compared with the index PN isolate), highly
2.3 III Indistinguishable PN 2 similar (1- or 2-fragment difference in PFGE profile), or unre-
2 0 III lated (.2-fragment difference). The proportion of all recurrent
2.6 III Indistinguishable PN 2
3 0 II
isolates available that represented a strain previously encoun-
5.8 II Indistinguishable PN 2 tered (indistinguishable or highly similar in PFGE) in the same
4 0 II infant was 65%. Our results are in accordance with 2 studies of
1.6 II Indistinguishable C1 women that have used PFGE and shown that 50%– 68% of
5 0 N
strains causing recurrent symptomatic cystitis were similar to
6.1 N Indistinguishable C1
6 0 II the index strains [13, 14]. Both studies included women in child-
3.5 II Indistinguishable C1 bearing age from North America. In one study, no woman had a
7 0 N history of UTI, and in the other all had a history of recurrent
2.4 NA NA C1 cystitis.
1.1 N Indistinguishable C2
Somewhat different results, however, have been published in
8 0 II
5.0 II Highly similar (1-fragment difference) C1 Scandinavia when other typing methods and different entry cri-
1.0 II Highly similar (1-fragment difference) C2 teria have been used. Namely, when Kärkkäinen et al. [15] used
9 0 III amplified polymorphic DNA PCR, only 25% of the recurrent
4.9 III Highly similar (1-fragment difference) C1 strains were indistinguishable when compared with the original
10 0 N
PN strains. In addition, our results are in disagreement with re-
3.0 N Highly similar (2-fragment difference) PN 2
11 0 II sults of a study of adults by Brauner et al. [16] in which only 33%
2.8 II Highly similar (2-fragment difference) C1 of the recurrences were caused by strains indistinguishable from
12 0 II the original PN strain. Their study was based on a biochemical
3.5 II Unrelated C1 phenotypic method, which is usually less sensitive than geno-
0.8 II Indistinguishable C2
13 0 III
typing by PFGE in distinguishing strains of E. coli. When ribo-
0.8 III Unrelated C1 typing was used in the study by Ikäheimo et al. [17], 33% of the
14 0 II strains isolated from recurrent infections were indistinguishable
1.6 NA NA C1 from the original strain after an index episode of cystitis. All the
3.6 II Unrelated PN 2 above-mentioned studies [15–17] included women in fertile age
15 0 II
4.5 N Unrelated C1
but also postmenopausal women with a history of UTI. In 2 of
2.1 III Unrelatedc C2 the studies [16, 17] the index UTI was PN, and in 1 study [15]
1.0 III Unrelatedc C3 the index UTI was cystitis. All the isolates from recurrent
16 0 N UTIs were isolates from patients with both symptomatic and
2.2 N Unrelated C1
asymptomatic infections.
2.8 N Unrelatedc C2
0.8 N Unrelatedc C3
Young infants have several risk factors for recurrent infections.
17 0 N In addition to the virulence factors of the causative agents, com-
3.6 II NT PN 2 mon urinary tract abnormalities as well as functional constipation
9.3 NA NA C1 and impaired bladder emptying have been related to recurrent
NOTE. C, cystitis; N, papG negative; NA, not available; NT, nontypeable UTIs in children [18, 19]. Genetically indistinguishable isolates
by PFGE. of E. coli causing PN can be found in the child’s feces [20]. Chil-
a
PFGE typing is in reference to the index PN isolate from the same infant.
“Unrelated” isolates were different from the index isolate by .2 fragments. dren prone to UTIs carry P-fimbriated E. coli in their intestine
b
Number indicates recurrent UTI count. more often than do healthy children [21], but it has not been stud-
c
2d and 3d recurrent cystitis caused by isolates unrelated to index isolate but ied whether these P-fimbriated strains also cause recurrent UTIs.
identical to each other.
In the present study, it was of interest that 5 of the “first recur-
rence” episodes were PN, which shows the pyelonephritogenic
of some infants. There was no statistically significant difference potential of the strains or the persistence of an established upper
with regard to the total use of antimicrobials between infants urinary tract focus in certain subjects or both. Another interesting
who were infected by indistinguishable or highly similar isolates finding was that most recurrent infections, even of the same strain,
and those infected by unrelated isolates. manifested themselves as cystitis, not PN. This may demonstrate

Downloaded from https://academic.oup.com/jid/article-abstract/185/3/375/895950

by guest
on 07 June 2018
 378 Jantunen et al. JID 2002;185 (1 February)

Figure 1. Pulsed-field gel electrophoresis banding patterns of index pyelonephritis isolate and recurrent urinary isolates from 3 infants with
Escherichia coli pyelonephritis. Lanes 1– 3, Isolates from infant no. 12; lanes 4– 6, isolates from infant no. 8; lanes 7– 10, isolates from infant
no.16; lanes St, molecular size markers.

the pathogenic versatility of these strains and argue against a cat- References
egorical distinction between “pyelonephritogenic” versus “cysti-
1. Jodal U. The natural history of bacteriuria in childhood. Infect Dis Clin
togenic” pathotypes.
North Am 1987; 1:713–29.
In adults, it has been disputed whether a recurrent UTI is a re- 2. McKerrow W, Davidson-Lamb N, Jones PF. Urinary tract infection in
infection caused by a new organism rather than a relapse caused children. Br Med J (Clin Res Ed) 1984; 289:299–303.
by the index strain that has colonized either the gut or the peri- 3. Panaretto KS, Craig JC, Knight JF, Howman-Giles R, Sureshkumar P,
urethral area. Recent studies have demonstrated, however, that Roy LP. Risk factors for recurrent urinary tract infection in preschool
uropathogens can persist within the bladder tissue in underlying children. J Paediatr Child Health 1999; 35:454– 9.
4. Jakobsson B, Berg U, Svensson L. Renal scarring after acute pyelonephri-
epithelial cells and serve as a source for recurrent UTIs [22]. In
tis. Arch Dis Child 1994; 70:111–5.
future studies, serial rectal cultures are needed to study whether 5. Lomberg H, Hellström M, Jodal U, Leffler H, Lincoln K, Svanborg Eden
the recurrent UTIs are endogenous relapses from the colonic C. Virulence-associated traits in Escherichia coli causing first and re-
flora rather than reinfections caused by new strains or by relapses current episodes of urinary tract infection in children with or without
of pathogens persisting in the underlying epithelial layers of vesicoureteral reflux. J Infect Dis 1984; 150:561–9.
6. Marklund BI, Tennent JM, Garcia E, et al. Horizontal gene transfer of the
the bladder tissue. The present study design was a multicenter
Escherichia coli pap and prs pili operons as a mechanism for the devel-
follow-up study carried out in 5 pediatric hospitals. We feel opment of tissue-specific adhesive properties. Mol Microbiol 1992; 6:
that, although the number of strains collected and analyzed from 2225–42.
patients with recurrent UTIs was not very high (40 isolates), it 7. Johanson IM, Plos K, Marklund BI, Svanborg C. Pap, papG and prsG
is high enough to justify our conclusions. Hence, recurrent com- DNA sequences in Escherichia coli from the fecal flora and the urinary
munity-acquired UTIs after an initial PN are rarely infections tract. Microb Pathog 1993; 15:121– 9.
8. Otto G, Sandberg T, Marklund BI, Ulleryd P, Svanborg C. Virulence factors
caused by a novel strain. and pap genotype in Escherichia coli isolates from women with acute
pyelonephritis, with or without bacteremia. Clin Infect Dis 1993; 17:
448–56.
9. Jantunen ME, Siitonen A, Koskimies O, et al. Predominance of class II
Acknowledgments papG allele of Escherichia coli in pyelonephritis in infants with normal
urinary tract anatomy. J Infect Dis 2000; 181:1822–4.
We thank Joanna Koort, Ulla-Maija Nakari, Tarja Heiskanen,
10. Farmer JJ III. Enterobacteriaceae: introduction and identification. In:
Liisa Immonen, Susanna Lukinmaa, and Ritva Taipalinen, for ex-
Murray PR, Baron EJ, Pfaller MA, Tenover FC, Yolken RH, eds. Manual
cellent technical assistance, and Seppo Sarna for statistical advice. of clinical microbiology. 6th ed. Washington, DC: ASM Press, 1995:
We also thank the staff members in the participating hospitals: 438–50.
Aurora Hospital (Helsinki); Hospital for Children and Adolescents, 11. Kärkkäinen UM, Kauppinen J, Ikäheimo R, Katila ML, Siitonen A. Rapid
University of Helsinki (Helsinki); Jorvi Hospital (Espoo, Finland); and specific detection of three different G adhesin classes of P-fimbriae
Päijät-Häme Central Hospital (Lahti, Finland); and the Department in uropathogenic Escherichia coli by polymerase chain reaction. J Mi-
of Pediatrics at Tampere University Hospital (Tampere, Finland). crobiol Methods 1998; 34:23– 9.

Downloaded from https://academic.oup.com/jid/article-abstract/185/3/375/895950

by guest
on 07 June 2018
 JID 2002;185 (1 February) Recurrent Urinary Tract Infections 379

12. Lukinmaa S, Schildt R, Rinttilä T, Siitonen A. Salmonella enteritidis 17. Ikäheimo R, Siitonen A, Heiskanen T, et al. Recurrence of urinary tract in-
phage types 1 and 4: pheno- and genotypic epidemiology of recent out- fection in a primary care setting: analysis of a 1-year follow-up of 179
breaks in Finland. J Clin Microbiol 1999; 37:2176–82. women. Clin Infect Dis 1996; 22:91–9.
13. Foxman B, Zhang L, Tallman P, et al. Virulence characteristics of Escheri- 18. Blethyn AJ, Jenkins HR, Roberts R, Verrier Jones K. Radiological evidence
chia coli causing first urinary tract infection predict risk of second infec- of constipation in urinary tract infection. Arch Dis Child 1995; 73:534–5.
tion. J Infect Dis 1995; 172:1536–41. 19. Koff SA, Wagner TT, Jayanthi VR. The relationship among dysfunctional
14. Russo TA, Stapleton A, Wenderoth S, Hooton TM, Stamm WE. Chromo- elimination syndromes, primary vesicoureteral reflux and urinary tract
somal restriction fragment length polymorphism analysis of Escherichia infections in children. J Urol 1998; 160:1019– 22.
coli strains causing recurrent urinary tract infections in young women. J 20. Jantunen ME, Saxén H, Lukinmaa S, Ala-Houhala M, Siitonen A. Genomic
Infect Dis 1995; 172:440–5. identity of pyelonephritogenic Escherichia coli isolated from the blood,
15. Kärkkäinen UM, Ikäheimo R, Katila ML, Siitonen A. Recurrence of uri- urineandfecesofchildrenwithurosepsis.JMedMicrobiol2001; 50:650–2.
nary tract infections in adult patients with community-acquired pyelo- 21. Plos K, Connell H, Jodal U, et al. Intestinal carriage of P fimbriated
nephritis caused by E. coli: a 1-year follow-up. Scand J Infect Dis Escherichia coli and the susceptibility to urinary tract infection in
2000; 32:495–9. young children. J Infect Dis 1995; 171:625– 31.
16. Brauner A, Jacobson SH, Kuhn I. Urinary Escherichia coli causing recur- 22. Mulvey MA, Schilling JD, Martinez JJ, Hultgren SJ. Bad bugs and belea-
rent infections—a prospective follow-up of biochemical phenotypes. gured bladders: interplay between uropathogenic Escherichia coli and
Clin Nephrol 1992; 38:318–23. innate host defenses. Proc Natl Acad Sci USA 2000; 97:8829–35.

Downloaded from https://academic.oup.com/jid/article-abstract/185/3/375/895950

by guest
on 07 June 2018