Journal of Hospital Infection (1992) 22, 117-127
Virulence factors tract infections J. Benton*,
in EscherM2ia coli from urinary in patients with spinal injuries
J. Chawlaf,
S. Parryl
and D. Stickler*
*School of Pure and Applied Biology, University of Wales College of Cardiff, Cardiff CF13TL, UK, TWelsh Spinal Injuries Unit, Rookwood Hospital, Cardig CF5 2 YN, UK and $ Unilever Research Ltd, Sharnbrook, Bedford MK44 ILQ, UK Accepted for publication
14 July 1992
Summary:
A collection of 70 strains of Escherichia coli from urinary tract infections in spine-injured patients undergoing long-term bladder catheterization were tested for characteristics that have been associated with the ability to produce pyelonephritis. The incidence of the virulence factors were: mannose-resistant haemagglutinins (30%), P-fimbriae (17%), haemolysin (27%), K-antigens (28%) and aerobactin (by bioassay 33%, by gene probe 39%). Only 54% of the strains belonged to the 0-serotypes usually associated with urinary tract infections. E. coli carrying the full complement of virulence factors were rare in the urinary tract of the spinal patients and were not associated with episodes of symptomatic pyelonephritis. It is clear that the neuropathic bladder and the presence of the catheter permits a wide variety of bacterial types to colonize the urinary tract and cause infection of the kidney. The identification of host markers rather than bacterial factors is suggested as a more fruitful approach to the early detection of cases likely to progress to pyelonephritis in this group of patients.
Keywords: Escherichia coli; virulence catheterization;
spine-injured
factors; patients.
urinary
tract infections;
bladder
Introduction The prevention and treatment of urinary tract infections are major concerns for those involved in the care of patients with spinal injuries. The requirement for long-term management of the neuropathic bladder by intermittent or indwelling catheterization results in most patients having their urine recurrently or continuously infected with a diverse bacterial flora.lm3 Bacterial colonization of the urinary tract begins soon after injury and eventually complex mixed species of nosocomial organisms accumulate in the bladder.4-6 The spread of these infections to the upper urinary tract poses a major threat to the health of the paraplegic patient.7 The role of antimicrobial agents in the prevention and treatment of these infections has not been adequately defined.8,9 In many spinal units, Correspondence 0195-6701/92/100117+
to: Dr D. Stickler. c> 1992 The Hospital
II SOS.OO/O
117
Infection
Society
118
J. Benton et al.
pressures to limit the use of antibiotics because of the threat of the emergence of resistant organisms, has led to withholding antibiotic therapy until clinical symptoms indicate that the infection has reached the kidneys or the bloodstream.‘O’ll Uncertainty over which organism, in the mixed bacterial cultures commonly obtained from the urine, to target with chemotherapy, often leads to treatment with broad-spectrum antibiotics such as gentamicin. ‘i Earlier intervention with antibiotics specifically aimed at invasive pyelonephritic strains of bacteria as soon as they appear in the urine might be more effective in preserving renal integrity and reduce the number of episodes of symptomatic infection suffered by these patients. Such an approach would require the identification of host or bacterial markers to aid early diagnosis. Evidence from epidemiological studies, and experimental work with animal models, suggest that in the non-catheterized urinary tract the most important cause of pyelonephritis is a distinct group of E. coli 0-serotypes possessing a collection of attributes which facilitate renal disease.‘2-‘5 These virulence factors include adhesins such as the mannose-resistant haemagglutinins (MRHA) which bring about the adherence and colonization of the epithelial surfaces of the urinary tract, a-haemolysin and the iron sequestering siderophore aerobactin, which provide the pathogens with a supply of iron, and the capsular K-antigens which endow cells with an ability to resist the bactericidal action of normal human serum. The objectives of the work reported here were to establish the incidence of recovery of potentially pyelonephritic strains of E. coli in urinary infections of spine-injured patients and to observe whether these ‘virulent’ E. coli are associated with episodes of clinical pyelonephritis in these patients. Materials
and methods
Bacterial isolation The clinical isolates of bacteria used in this investigation were all obtained from the urine samples of patients with spinal injuries. Dr H. Green (Southport General Infirmary) and Dr P. Gillette (Stoke Mandeville Hospital) kindly provided 37 strains of E. coli; all others were obtained from Rookwood Hospital, Cardiff and the Cardiff Royal Infirmary. Urines were collected from indwelling catheters by clamping and aspiration with a syringe. In patients undergoing intermittent catheterization the urine was collected directly from the catheter outlet during mid-flow. When external condom drainage was being used for bladder management, the condom was removed, the urethral meatus cleaned with water and a mid-stream sample obtained by bladder expression. Urines were cultured onto CLED agar (Oxoid Ltd, UK), Pseudomonas Selective Medium (Oxoid Ltd) and CI agar.’
Urinary
infections
in spinal
patients
Isolates were identified using an appropriate kit (API Ltd). Cultures stored in a cryopreservative fluid at -70°C on plastic beads (TSC UK). Unless otherwise stated they were sub-cultured on to CLED prior to use. 0 and K serotyping The 0 and K serotypes agglutination tests using 07, 08, 09, 011, 015, kindly performed by Dr
119
were Ltd, agar
of the E. coli isolates were established by direct standard antisera to the types 01, 02, 04, 0.5, 06, 017, 018, 025, 075, Kl and KS. The typing was A. Roberts (Charing Cross Hospital, London).
Detection of haemolysin Washed sheep red blood cells were added to molten (45°C) Blood Agar Base No. 2 (Oxoid Ltd) to a final concentration of 5% w/v. The plates were inoculated with the test strains and incubated overnight at 37°C. Haemolysin production was indicated by a zone of clearing around the colonies. The strains were sub-cultured on the blood agar daily for 7 days and were considered to be haemolytic only if they showed activity on 2 consecutive days.r4 Detection of aerobactin The bioassay of aerobactin was performed by the method of Carbonetti and Williams.” A 2 kb AvaI fragment within the aerobactin synthesis gene and a 2.3 kb PvuII fragment from within the aerobactin receptor gene, kindly provided by Dr P. H. Williams of Leicester University were used to probe the strains for the genes of the aerobactin system. The colony hybridization technique described by Carbonetti et a1.‘7 was used. Haemagglutination The method used was a modification of that devised by Duguid et a1.18 The E. coli isolates were grown overnight at 37°C on Brain Heart Infusion agar (Oxoid) to give optimal expression of mannose-resistant (MR) fimbriae and in Brain Heart Infusion broth (Oxoid) for optimal expression of mannose-sensitive (MS) fimbriae. The isolates were serially subcultured twice on the agar and three times in the broth. Bacteria were then harvested, washed and resuspended in phosphate buffered saline (PBS) (pH 7.3) to an OD,,, of 2.0 (c. lo9 cfu ml-‘). Aliquots (50 ~1) of these test suspensions were added to duplicate wells on glazed white porcelain tiles, and either 50 ~1 of sterile PBS or 50 l.tl of 30/o ( w / v ) D - mannose (Sigma Ltd) in PBS was added to each pair of wells. Finally, 50 ~1 of either guinea-pig or human type 0 erythrocytes resuspended to 4% ( v / v ) in PBS were added to the wells. Inhibition of agglutination of guinea-pig erythrocytes by mannose was considered indicative of the presence of MS fimbriae and agglutination of human erythrocytes in the presence of mannose indicated possession of MR fimbriae.
120
J. Benton
et al.
P-fimbriae Latex particles coated in &n-Gal-( I-4)-P-D-Gal disaccharide (Orion Diagnostics, Finland) were used to detect P-fimbriae. Isolates were prepared by growth overnight at 37°C on Tryptone Soya agar (Oxoid). Four bacterial colonies were harvested from each agar plate and resuspended in a drop of sterile PBS, the cells were then mixed with pre-dried latex particles on a test card and the agglutination reaction read after 2 min. A negative control (dried latex particles without Gal-Gal) was included alongside each sample. In addition, a positive control (a pre-dried P+ strain together with Gal-Gal latex spheres) was included with each set of tests.
Statistical analysis The significance of differences between virulence factors expressed by E. coli with urinary 0-serotypes, non-urinary serotypes and autoagglutinating strains was examined by the Chi-squared test. Results
The prevalence of various bacterial species in the urine of catheterized patients Over an l&month period 820 urine samples were collected from 32 The majority of the samples (68%) revealed spine-injured patients. significant bacteriuria ( > lo5 cfu ml-‘). The prevalence of the various bacterial species in these specimens is shown in Table I. Enterococcus faecalis, Klebsiella pneumoniae, E. coli and Proteus mirabilis were the predominant organisms; E. coli represented 14.5% of the total isolates. Table I. The incidence of bacterial species in the urine of 32 spine-injured patients Organism Enterococcus faecalis Klebsiella pneumoniae Escherichia coli Proteus mirabilis Acinetobacter calcoaceticus Pseudomonas aeruginosa Enterobacter cloacae Staphylococcus epidermidis Staphylococcus aureus Providencia stuartii Staphylococcus haemolyticus Staphylococcus xylosus Citrobacter diversus Micrococcus spp. Candida albicans Enterobacter aerogenes Others
No. of isolates WJ) 208 (17%) 200 (17.1) 169 (14.5) 110 ( 9.4) 92 ( 7.9) 67 ( 5.7) 62 ( 5.3) 45 ( 3.8) 41 ( 3.5) 41 ( 3.5) 33 ( 2.8) 24 ( 2.1) 18 ( 1.5) 16 ( 1.4) 9 ( 0.8) 7 ( 0.6) 27 ( 2.3)
Urinary
infections
in spinal
patients
121
Characterization
of the Escherichia coli isolates From the 169 isolates of E. coli obtained from 13 of the Cardiff patients, 70 were selected for further analysis. All those isolated on a single occasion from a particular patient were included. In those cases where patients had long-standing E. coli infections, representative strains from each week were included, together with any strains showing changed antibiotic sensitivity patterns or API biotypes. In addition, every isolate recovered during an episode of clinical pyelonephritis was selected. These 70 isolates were tested for 0-serotype, Kl and KS antigens, haemolysin, MSHA, MRSA, P-fimbriae and aerobactin. Repeated isolates from the same patient showing identical 0-serotype, virulence profile and API biotype were then excluded from the collection. The prevalence of the virulence factors in the resulting 33 Cardiff strains and those obtained from Southport and Stoke Mandeville are presented in Table II. Thirty-eight (54%) of these 70 strains were members of 0-serotypes normally associated with urinary tract infection. Chi-squared analysis revealed a significant difference (PC 0.01) between the prevalence of virulence factors in these serotypes and the non-urinary serotypes and autoagglutinating strains, haemolysin, MRHA and P-fimbriae being more prevalent in the urinary serotypes. None of the strains, however, produced the full battery of the five virulence factors (0-serotype, K antigens, MRHA, haemolysin and aerobactin): 8.6% produced four of the factors, 22.9% produced three, 22.9% produced two, 28.6% produced only one and 17% produced none. A comparison of the frequency of virulence markers in E. coli isolated in this study and those reported from surveys on other groups of patients is presented in Table III. The prevalence of P-fimbriae and other adhesins responsible for MRHA, haemolysin and aerobactin in E. coli causing urinary tract infections in spine-injured patients was much lower than that reported in strains causing pyelonephritis in non-catheterized women and children.
A prospective study of the association of urinary organisms with episodes of clinical pyelonephritis During the course of the study 22 episodes were recorded where patients were diagnosed by a clinician, on the basis of clinical symptoms, as suffering from pyelonephritis and where a recent sample of urine was available for analysis. The organisms isolated from these urines are listed in Table IV. A variety of bacterial species were isolated from the urine on these occasions. E. coli was the most common, but 10 other species were also isolated. E. coli associated with bouts of pyelonephritis ranged from a strain which did not express a urinary 0-serotype and was devoid of all virulence characteristics, to an 04 serotype which was capable of producing MSHA, MRHA, P-fimbriae, and haemolysin (Table V).
W4)
2(9)
38
22 10 70
13(18)
2(20)
4
1
2(9) 3(30) 7(10)
Z(5)
2
0: i
0: :
z i
8
:
A
:
04 06 :i 01.5
018
0
2
4
00:
075 Total urinary serotypes Non-urinary serotypes Autoagglutinating Total
KS antigen
K, antigen
2i 3 1 4
i : 0 5
19(27)
2(9)
2(W
16(73) 4(40) 52(74)
31(82)
4
4
lS(39)
43
MSHA
01
Haemolysin
possessing
2
2
12(17)
WQ 21(30)
l(5)
WO)
4(18)
9(26)
:
0
lS(39)
:,
:
;
P-fimbriae
; :,
3
;
MRHA
each characteristic
4(18) VW 23(33)
14(37)
; 2 1 2
:
:
Bioassay
(W)
and virulence markers of the E. coli isolates
No. of isolates
II. 0-serotypes
No. of isolates
0-serotypes
Table
WO) 27(39)
4(18)
28(40)
17(45)
i
: 1 S(39)
: 1 3
i 1
Receptor gene
i i
1 2 1
Synthesis gene
Aerobactin
k
!t
f
if
?
Urinary
infections
in spinal
patients
123
124
J. Benton
et al.
Table IV. Organisms associated with the onset of clinical symptoms of pyelonephritis Patient No. 413 416 470 484 484 513 522 522 550 550 550 551 5.51 555 55.5 605 650 6.50 660 669 669 790
Day of onset of symptoms 6 17 85 1 27 53 7 22 1 27 107 1 23 35 110 12 10:: 17 1 It
Day of urine sample
1: 86 2: 51 2;: 1 1:; 2 23 32 102 10 10 104 19 1 10 26
Organism(s) isolated from urine Pr. mirabilis and A. calcoaceticus E. coli and E. faecalis K. pneumoniae K. pneumoniae E. faecalis E. cloacae E. faecalis E. coli and E. faecalis S. aureus E. coli E. coli E. faecalis P. aeruginosa, C. diversus and Pr. mirabilis E. coli and E. aerogenes K. pneumoniae and Pr. stuartii A. calcoaceticus Pr . mirabilis E. cola’ E. coli and Pr. stuartii E. coli E. coli Pr. mirabilis
Discussion
The variety of organisms isolated in this study (Table I) is similar to that reported in previous prospective studies of urinary tract infections in and in elderly patients undergoing long-term spine-injured patients’-’ indwelling catheterization.27 E. coli is not the major cause of infection in these patients; it is just one of a range of species that comprise the dynamic Table V. The prevalence of virulence factors in the strains of Escherichia coli isolated at the onset of pyelonephritis Patient No. Virulence profile of isolates 416 522 550(a) 550(b) 555 650 660 669(a) 669(b)
07; NT: 04;’ 04; 06; 06; NT; 075; NT;
KU; KU: KU; KU; KU; KU; KU;
MS+; MS+: MS+; MS+; MS+; MS+; MS-;
MR+ MRMR+ MR+ MRMR+ MR-
Kl;
MS+;
MR-
KU; MS-;
MR-
P-; P-; P+; P+; P-; P-: P-; P-; P-;
H-; H-; H+; H+; H+; H+: H-f H+; H-;
A+. AA-: A-. AA-: A+. A-. A-.
NT=not typable with urinary O-antisera, KU = not typable with Kl or KS antisera, MS=mannose-sensitive haemagglutinin, MR = mannose-resistant haemagglutinin, P = P-fimbriae, H = haemolysin, A = aerobactin. + = Presence of factor, - = absence of factor.
Urinary
infections
in spinal
patients
125
and often polymicrobial communities that colonize the catheterized urinary tract. It is also obvious from Table II that strains of E. coli possessing the full battery of virulence markers are rarely isolated from the urine of spine-injured patients. There was a higher incidence of virulence factors in the 0-serotypes generally associated with urinary tract infections, than in the autoagglutinable strains and the non-urinary 0-serotypes, but overall the occurrence of these factors was much lower than in collections of E. coli from cases of pyelonephritis in non-catheterized patients (Table III). Only 17% of the strains from the spinal patients produced P-fimbriae, an incidence similar to that reported for normal faecal isolates (18%) and in marked contrast to the 80% recorded for pyelonephritic E. coli.23,25*26It is interesting that while 21 of the strains were capable of expressing MRHA, nine of these did not produce P-fimbriae, suggesting that these particular strains were producing other types of haemagglutinins, the X adhesins28 which have different receptor specificity and tissue tropism from the P-fimbriae.29 The results with aerobactin and haemolysin (Table III) suggest that these systems which provide iron for the growth of pyelonephritic strains’6,30 are not necessary for the stable colonization of the catheterized tract. Mobley et ~2.~’ reported that most E. coli isolated from the urine of catheterized geriatric patients expressed MSHA. The proportions of isolates from short (1 week), medium (2-11 weeks) and long (> 12 weeks) episodes of bacteriuria which expressed MSHA was reported as 59, 65 and 92% respectively. These authors therefore proposed that the ability of E. coli to persist in the catheterized urinary trrlct may depend on the expression of the type 1 fimbriae responsible for MSHA and that other virulence factors may not be required for persistent colonization and pathogenesis in these circumstances. Our study provides some support for this hypothesis in that 76% of our isolates produced MSHA (Table III). Analysis of the 28 defined episodes of E. coli bacteriuria showed that the six most stable colonizers that produced episodes which lasted for 4 weeks or more, were all MS+. In addition it was observed that MS+MRstrains were capable of colonizing the catheterized tract for up to 35 days. Lomberg et a1.21 noted that in young girls with vesicoureteral reflux, pyelonephritic infections were often associated with organisms other than E. cob and that infecting strains of this species rarely carried the virulence factors that were found in the strains producing pyelonephritis in girls who did not have reflux. Vesicoureteral reflux can be common in some spine-injured patients.3’ The data presented in Table IV shows that E. coli was present in only nine of the 22 incidents in which pyelonephritis was diagnosed on the basis of clinical symptoms (fever, nausea, rigors) during the study. It is also clear that these E. coli were poorly endowed with virulence factors (Table V). In conclusion it seems that virulence factors indicative of pyelonephritic strains cannot be used to predict that an E. coli infection in these patients
126
J. Benton
et al.
will lead to an episode of symptomatic pyelonephritis. The neuropathic bladder and the catheterization process allows a variety of organisms that are not virulent for the normal urinary tract to persist in the bladder urine and initiate pyelonephritis. It thus seems that in this compromised group of patients, investigation of host rather than bacterial markers might be a more fruitful approach to the early detection of those cases likely to progress to pyelonephritis. References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.
12.
13.
14.
Newman E, Price M. Bacteriuria in patients with spinal cord lesions: its relationship to urinary drainage appliances. Arch Phys Med Rehabil 1977; 58: 427430. Donovan WH, Stolov WC, Clowers DE, Clowers MR. Bacteriuria during intermittent catheterization following spinal cord injury. Arch Phys Med Rehabill978; 59: 35 l-3.57. Erikson RP, Merritt JL, Opitz JL, Istrup DM. Bacteriuria during follow-up in patients with spinal cord injury. Arch Phys Med Rehabil 1982; 63: 409412. Stickler DJ, Wilmot CB, O’Flynn JD. The mode of development of urinary infection in intermittently catheterized male paraplegics. Paraplegia 1971; 8: 243-252. Clayton CL, Chawla JC, Stickler DJ. Some observations on urinary tract infections in patients undergoing long-term bladder catheterization. J Hasp Infect 1982; 3: 3947. Fawcett C, Chawla JC, Quoraishi A, Stickler DJ. A study of the skin flora of spinal cord injured patients. J Hosp Infect 1986; 8: 149-158. Slade N, Gillespie WA. The Urinary Tract and the Catheter. New York: John Wiley 1985; 49. Warren JW, Muncie HL, Bergquist EJ, Hoopes JM. Sequelae and management of urinary infection in the patient requiring chronic catheterization.? Ural 1981; 125: l-8. Stickler DJ, Chawla JC. The role of antiseptics in the management of patients with long-term indwelling bladder catheters. J Hasp Infect 1987; 10: 219-228. Galloway A, Green HT, Windsor JJ, Menon KK, Gardner BD. Serial concentrations of C-reactive protein as an indicator of urinary tract infection in patients with spinal injury. J Clin Path 1986; 39: 851-855. Stickler DJ, Chawla JC. An appraisal of antibiotic policies for urinary tract infections in patients with spinal cord injuries undergoing long-term intermittent catheterization. Paraplegia 1988; 26: 215-225. Hagberg L, Jodal U, Korhonen TK, Lidin-Janson G, Lindberg U, Svanborg-Eden C. Adhesion, haemagglutination and virulence of Escherichia coli causing urinary tract infections. Infect Immun 1981; 31: 564-570. Winberg J. Balance between host defence and bacterial virulence in urinary tract infection. In: Asscher AW, Brumfitt W, Eds. Microbial Diseases in Nephrology. New York: John Wiley 1986; 197-223. O’Hanley P, Low D, Romero I, Lark D, Vosti K, Falkow S, Schoolnik G. Gal-Gal binding and haemolysin phenotypes associated with uropathogenic Escherichia coli. N
EnglJ Med 1985; 313: 414-420. 15. Johnson JR. Virulence factors in Escherichia coli urinary tract infection. Clin Microbial Revs 1991; 4: 86128. NH, Williams PH. Detection of synthesis of the hydroxamate siderophore 16. Carbonetti aerobactin by pathogenic isolates of Escherichia coli. In: Sussman M, Ed. The Virulence of the Society for General of Escherichia coli: Reviews and Methods. Special publications Microbiology 13. London: Academic Press 1985; 419424. NH, Boonchai S, Parry SH, Vaisanen-Rhen V, Korhonen TK, Williams 17. Carbonetti PH. Aerobactin-mediated iron uptake by Escherichia coli isolates from human extra-intestinal infections. Infect Immun 1986; 51: 966-968. and non-fimbrial haemagglutinins of 18. Duguid JP, Clegg S, Wilson MI. The fimbrial Escherichia coli. J Med Microbial 1979; 12: 213-227. U, Hanson A, Jodal V, Lidin-Janson G, Lincoln K, Olling S. Asymptomatic 19. Lindberg bacteriuria in schoolgirls: differences in Escherichia coli causing asymptomatic and symptomatic bacteriuria. Acta Paediatr Band 1975; 64: 432436.
Urinary
infections
in spinal
patients
127
20. Mobley HLT, Chippendale GR, Tenney JH, Hull RA, Warren JW. Expression of type 1 fimbriae may be required for persistence of Escherichia coli in the catheterized urinary tract. 7 Clin Microbial 1987: 25: 2253-2257. 21. Lomb&g H, Hellstrom H, Jbdal V, Leffler H, Lincoln K, Svanborg-Eden C. Virulence associated traits in Escherichia coli causing first and recurrent episodes of urinary tract infection in children with and without vesicoureteral reflux. J Infect Dis 1984; 150: 561-569. 22. Lidin-Janson G, Hanson LA, Kaijser B et al. Comparison of Escherichia coli from bacteriuric patients with those from faeces of healthy schoolchildren. J Infect Dis 1977;
136: 346-353. 23. Kallenius G, Jacobson SH, Tullus K, Svenson SB. P-fimbriae studies on the diagnosis and prevention of acute pyelonephritis. Infection 1985; 13: 159-161. 24. Funfstuck R, Tschape H, Stein G, Kunath H, Bergner M, Wessel G. Virulence properties of Escherichia coli strains in patients with chronic pyelonephritis. Infection 1986; 14: 145-150. 25. Enerback S, Larsson AC, Leffler H et al. Binding to galactose a l-4 galactose containing receptors a potential tool in urinary tract infection. J Clin Microbial 1987; 25: 407-411. 26. Westerlund B, Sitonen A, Elo J, Williams PH, Korhonen TK, Makela PH. Properties of Escherichia coli isolates from urinary tract infections in boys. J Infect Dis 1988; 158: 996-1002. 27. Warren JW, Tenney JH, Hoopes JM, Muncie HL, Anthoney WC. A prospective microbiologic study of bacteriuria in patients with chronic indwelling uretheral catheters. r Infect Dis 1982; 146: 719-723. 28. Vaisanen-Rhen V. Fimbria-like haemagglutinin of Escherichia coli 075 strains. Infect Immun 1984; 46: 401-407. 29. Nowicki B, Truong L, Movids J, Hull R. Presence of the Dr receptor in normal human tissues and its possible role in the pathogenesis of ascending urinary tract infection. AmJ Path 1988; 133: 14. 30. Calveri SJ, Bohach G, Snyder IS. Escherichia coli o-haemolysin, characteristics and probable role in pathogenicity. Microbial Revs 1984; 48: 326-343. 31. Lamid S. Long-term follow-up of spinal cord injury patients with vesicoureteral reflux. Paraplegia 1988; 26: 27-34.
Virulence factors tract infections J. Benton*,
in EscherM2ia coli from urinary in patients with spinal injuries
J. Chawlaf,
S. Parryl
and D. Stickler*
*School of Pure and Applied Biology, University of Wales College of Cardiff, Cardiff CF13TL, UK, TWelsh Spinal Injuries Unit, Rookwood Hospital, Cardig CF5 2 YN, UK and $ Unilever Research Ltd, Sharnbrook, Bedford MK44 ILQ, UK Accepted for publication
14 July 1992
Summary:
A collection of 70 strains of Escherichia coli from urinary tract infections in spine-injured patients undergoing long-term bladder catheterization were tested for characteristics that have been associated with the ability to produce pyelonephritis. The incidence of the virulence factors were: mannose-resistant haemagglutinins (30%), P-fimbriae (17%), haemolysin (27%), K-antigens (28%) and aerobactin (by bioassay 33%, by gene probe 39%). Only 54% of the strains belonged to the 0-serotypes usually associated with urinary tract infections. E. coli carrying the full complement of virulence factors were rare in the urinary tract of the spinal patients and were not associated with episodes of symptomatic pyelonephritis. It is clear that the neuropathic bladder and the presence of the catheter permits a wide variety of bacterial types to colonize the urinary tract and cause infection of the kidney. The identification of host markers rather than bacterial factors is suggested as a more fruitful approach to the early detection of cases likely to progress to pyelonephritis in this group of patients.
Keywords: Escherichia coli; virulence catheterization;
spine-injured
factors; patients.
urinary
tract infections;
bladder
Introduction The prevention and treatment of urinary tract infections are major concerns for those involved in the care of patients with spinal injuries. The requirement for long-term management of the neuropathic bladder by intermittent or indwelling catheterization results in most patients having their urine recurrently or continuously infected with a diverse bacterial flora.lm3 Bacterial colonization of the urinary tract begins soon after injury and eventually complex mixed species of nosocomial organisms accumulate in the bladder.4-6 The spread of these infections to the upper urinary tract poses a major threat to the health of the paraplegic patient.7 The role of antimicrobial agents in the prevention and treatment of these infections has not been adequately defined.8,9 In many spinal units, Correspondence 0195-6701/92/100117+
to: Dr D. Stickler. c> 1992 The Hospital
II SOS.OO/O
117
Infection
Society
118
J. Benton et al.
pressures to limit the use of antibiotics because of the threat of the emergence of resistant organisms, has led to withholding antibiotic therapy until clinical symptoms indicate that the infection has reached the kidneys or the bloodstream.‘O’ll Uncertainty over which organism, in the mixed bacterial cultures commonly obtained from the urine, to target with chemotherapy, often leads to treatment with broad-spectrum antibiotics such as gentamicin. ‘i Earlier intervention with antibiotics specifically aimed at invasive pyelonephritic strains of bacteria as soon as they appear in the urine might be more effective in preserving renal integrity and reduce the number of episodes of symptomatic infection suffered by these patients. Such an approach would require the identification of host or bacterial markers to aid early diagnosis. Evidence from epidemiological studies, and experimental work with animal models, suggest that in the non-catheterized urinary tract the most important cause of pyelonephritis is a distinct group of E. coli 0-serotypes possessing a collection of attributes which facilitate renal disease.‘2-‘5 These virulence factors include adhesins such as the mannose-resistant haemagglutinins (MRHA) which bring about the adherence and colonization of the epithelial surfaces of the urinary tract, a-haemolysin and the iron sequestering siderophore aerobactin, which provide the pathogens with a supply of iron, and the capsular K-antigens which endow cells with an ability to resist the bactericidal action of normal human serum. The objectives of the work reported here were to establish the incidence of recovery of potentially pyelonephritic strains of E. coli in urinary infections of spine-injured patients and to observe whether these ‘virulent’ E. coli are associated with episodes of clinical pyelonephritis in these patients. Materials
and methods
Bacterial isolation The clinical isolates of bacteria used in this investigation were all obtained from the urine samples of patients with spinal injuries. Dr H. Green (Southport General Infirmary) and Dr P. Gillette (Stoke Mandeville Hospital) kindly provided 37 strains of E. coli; all others were obtained from Rookwood Hospital, Cardiff and the Cardiff Royal Infirmary. Urines were collected from indwelling catheters by clamping and aspiration with a syringe. In patients undergoing intermittent catheterization the urine was collected directly from the catheter outlet during mid-flow. When external condom drainage was being used for bladder management, the condom was removed, the urethral meatus cleaned with water and a mid-stream sample obtained by bladder expression. Urines were cultured onto CLED agar (Oxoid Ltd, UK), Pseudomonas Selective Medium (Oxoid Ltd) and CI agar.’
Urinary
infections
in spinal
patients
Isolates were identified using an appropriate kit (API Ltd). Cultures stored in a cryopreservative fluid at -70°C on plastic beads (TSC UK). Unless otherwise stated they were sub-cultured on to CLED prior to use. 0 and K serotyping The 0 and K serotypes agglutination tests using 07, 08, 09, 011, 015, kindly performed by Dr
119
were Ltd, agar
of the E. coli isolates were established by direct standard antisera to the types 01, 02, 04, 0.5, 06, 017, 018, 025, 075, Kl and KS. The typing was A. Roberts (Charing Cross Hospital, London).
Detection of haemolysin Washed sheep red blood cells were added to molten (45°C) Blood Agar Base No. 2 (Oxoid Ltd) to a final concentration of 5% w/v. The plates were inoculated with the test strains and incubated overnight at 37°C. Haemolysin production was indicated by a zone of clearing around the colonies. The strains were sub-cultured on the blood agar daily for 7 days and were considered to be haemolytic only if they showed activity on 2 consecutive days.r4 Detection of aerobactin The bioassay of aerobactin was performed by the method of Carbonetti and Williams.” A 2 kb AvaI fragment within the aerobactin synthesis gene and a 2.3 kb PvuII fragment from within the aerobactin receptor gene, kindly provided by Dr P. H. Williams of Leicester University were used to probe the strains for the genes of the aerobactin system. The colony hybridization technique described by Carbonetti et a1.‘7 was used. Haemagglutination The method used was a modification of that devised by Duguid et a1.18 The E. coli isolates were grown overnight at 37°C on Brain Heart Infusion agar (Oxoid) to give optimal expression of mannose-resistant (MR) fimbriae and in Brain Heart Infusion broth (Oxoid) for optimal expression of mannose-sensitive (MS) fimbriae. The isolates were serially subcultured twice on the agar and three times in the broth. Bacteria were then harvested, washed and resuspended in phosphate buffered saline (PBS) (pH 7.3) to an OD,,, of 2.0 (c. lo9 cfu ml-‘). Aliquots (50 ~1) of these test suspensions were added to duplicate wells on glazed white porcelain tiles, and either 50 ~1 of sterile PBS or 50 l.tl of 30/o ( w / v ) D - mannose (Sigma Ltd) in PBS was added to each pair of wells. Finally, 50 ~1 of either guinea-pig or human type 0 erythrocytes resuspended to 4% ( v / v ) in PBS were added to the wells. Inhibition of agglutination of guinea-pig erythrocytes by mannose was considered indicative of the presence of MS fimbriae and agglutination of human erythrocytes in the presence of mannose indicated possession of MR fimbriae.
120
J. Benton
et al.
P-fimbriae Latex particles coated in &n-Gal-( I-4)-P-D-Gal disaccharide (Orion Diagnostics, Finland) were used to detect P-fimbriae. Isolates were prepared by growth overnight at 37°C on Tryptone Soya agar (Oxoid). Four bacterial colonies were harvested from each agar plate and resuspended in a drop of sterile PBS, the cells were then mixed with pre-dried latex particles on a test card and the agglutination reaction read after 2 min. A negative control (dried latex particles without Gal-Gal) was included alongside each sample. In addition, a positive control (a pre-dried P+ strain together with Gal-Gal latex spheres) was included with each set of tests.
Statistical analysis The significance of differences between virulence factors expressed by E. coli with urinary 0-serotypes, non-urinary serotypes and autoagglutinating strains was examined by the Chi-squared test. Results
The prevalence of various bacterial species in the urine of catheterized patients Over an l&month period 820 urine samples were collected from 32 The majority of the samples (68%) revealed spine-injured patients. significant bacteriuria ( > lo5 cfu ml-‘). The prevalence of the various bacterial species in these specimens is shown in Table I. Enterococcus faecalis, Klebsiella pneumoniae, E. coli and Proteus mirabilis were the predominant organisms; E. coli represented 14.5% of the total isolates. Table I. The incidence of bacterial species in the urine of 32 spine-injured patients Organism Enterococcus faecalis Klebsiella pneumoniae Escherichia coli Proteus mirabilis Acinetobacter calcoaceticus Pseudomonas aeruginosa Enterobacter cloacae Staphylococcus epidermidis Staphylococcus aureus Providencia stuartii Staphylococcus haemolyticus Staphylococcus xylosus Citrobacter diversus Micrococcus spp. Candida albicans Enterobacter aerogenes Others
No. of isolates WJ) 208 (17%) 200 (17.1) 169 (14.5) 110 ( 9.4) 92 ( 7.9) 67 ( 5.7) 62 ( 5.3) 45 ( 3.8) 41 ( 3.5) 41 ( 3.5) 33 ( 2.8) 24 ( 2.1) 18 ( 1.5) 16 ( 1.4) 9 ( 0.8) 7 ( 0.6) 27 ( 2.3)
Urinary
infections
in spinal
patients
121
Characterization
of the Escherichia coli isolates From the 169 isolates of E. coli obtained from 13 of the Cardiff patients, 70 were selected for further analysis. All those isolated on a single occasion from a particular patient were included. In those cases where patients had long-standing E. coli infections, representative strains from each week were included, together with any strains showing changed antibiotic sensitivity patterns or API biotypes. In addition, every isolate recovered during an episode of clinical pyelonephritis was selected. These 70 isolates were tested for 0-serotype, Kl and KS antigens, haemolysin, MSHA, MRSA, P-fimbriae and aerobactin. Repeated isolates from the same patient showing identical 0-serotype, virulence profile and API biotype were then excluded from the collection. The prevalence of the virulence factors in the resulting 33 Cardiff strains and those obtained from Southport and Stoke Mandeville are presented in Table II. Thirty-eight (54%) of these 70 strains were members of 0-serotypes normally associated with urinary tract infection. Chi-squared analysis revealed a significant difference (PC 0.01) between the prevalence of virulence factors in these serotypes and the non-urinary serotypes and autoagglutinating strains, haemolysin, MRHA and P-fimbriae being more prevalent in the urinary serotypes. None of the strains, however, produced the full battery of the five virulence factors (0-serotype, K antigens, MRHA, haemolysin and aerobactin): 8.6% produced four of the factors, 22.9% produced three, 22.9% produced two, 28.6% produced only one and 17% produced none. A comparison of the frequency of virulence markers in E. coli isolated in this study and those reported from surveys on other groups of patients is presented in Table III. The prevalence of P-fimbriae and other adhesins responsible for MRHA, haemolysin and aerobactin in E. coli causing urinary tract infections in spine-injured patients was much lower than that reported in strains causing pyelonephritis in non-catheterized women and children.
A prospective study of the association of urinary organisms with episodes of clinical pyelonephritis During the course of the study 22 episodes were recorded where patients were diagnosed by a clinician, on the basis of clinical symptoms, as suffering from pyelonephritis and where a recent sample of urine was available for analysis. The organisms isolated from these urines are listed in Table IV. A variety of bacterial species were isolated from the urine on these occasions. E. coli was the most common, but 10 other species were also isolated. E. coli associated with bouts of pyelonephritis ranged from a strain which did not express a urinary 0-serotype and was devoid of all virulence characteristics, to an 04 serotype which was capable of producing MSHA, MRHA, P-fimbriae, and haemolysin (Table V).
W4)
2(9)
38
22 10 70
13(18)
2(20)
4
1
2(9) 3(30) 7(10)
Z(5)
2
0: i
0: :
z i
8
:
A
:
04 06 :i 01.5
018
0
2
4
00:
075 Total urinary serotypes Non-urinary serotypes Autoagglutinating Total
KS antigen
K, antigen
2i 3 1 4
i : 0 5
19(27)
2(9)
2(W
16(73) 4(40) 52(74)
31(82)
4
4
lS(39)
43
MSHA
01
Haemolysin
possessing
2
2
12(17)
WQ 21(30)
l(5)
WO)
4(18)
9(26)
:
0
lS(39)
:,
:
;
P-fimbriae
; :,
3
;
MRHA
each characteristic
4(18) VW 23(33)
14(37)
; 2 1 2
:
:
Bioassay
(W)
and virulence markers of the E. coli isolates
No. of isolates
II. 0-serotypes
No. of isolates
0-serotypes
Table
WO) 27(39)
4(18)
28(40)
17(45)
i
: 1 S(39)
: 1 3
i 1
Receptor gene
i i
1 2 1
Synthesis gene
Aerobactin
k
!t
f
if
?
Urinary
infections
in spinal
patients
123
124
J. Benton
et al.
Table IV. Organisms associated with the onset of clinical symptoms of pyelonephritis Patient No. 413 416 470 484 484 513 522 522 550 550 550 551 5.51 555 55.5 605 650 6.50 660 669 669 790
Day of onset of symptoms 6 17 85 1 27 53 7 22 1 27 107 1 23 35 110 12 10:: 17 1 It
Day of urine sample
1: 86 2: 51 2;: 1 1:; 2 23 32 102 10 10 104 19 1 10 26
Organism(s) isolated from urine Pr. mirabilis and A. calcoaceticus E. coli and E. faecalis K. pneumoniae K. pneumoniae E. faecalis E. cloacae E. faecalis E. coli and E. faecalis S. aureus E. coli E. coli E. faecalis P. aeruginosa, C. diversus and Pr. mirabilis E. coli and E. aerogenes K. pneumoniae and Pr. stuartii A. calcoaceticus Pr . mirabilis E. cola’ E. coli and Pr. stuartii E. coli E. coli Pr. mirabilis
Discussion
The variety of organisms isolated in this study (Table I) is similar to that reported in previous prospective studies of urinary tract infections in and in elderly patients undergoing long-term spine-injured patients’-’ indwelling catheterization.27 E. coli is not the major cause of infection in these patients; it is just one of a range of species that comprise the dynamic Table V. The prevalence of virulence factors in the strains of Escherichia coli isolated at the onset of pyelonephritis Patient No. Virulence profile of isolates 416 522 550(a) 550(b) 555 650 660 669(a) 669(b)
07; NT: 04;’ 04; 06; 06; NT; 075; NT;
KU; KU: KU; KU; KU; KU; KU;
MS+; MS+: MS+; MS+; MS+; MS+; MS-;
MR+ MRMR+ MR+ MRMR+ MR-
Kl;
MS+;
MR-
KU; MS-;
MR-
P-; P-; P+; P+; P-; P-: P-; P-; P-;
H-; H-; H+; H+; H+; H+: H-f H+; H-;
A+. AA-: A-. AA-: A+. A-. A-.
NT=not typable with urinary O-antisera, KU = not typable with Kl or KS antisera, MS=mannose-sensitive haemagglutinin, MR = mannose-resistant haemagglutinin, P = P-fimbriae, H = haemolysin, A = aerobactin. + = Presence of factor, - = absence of factor.
Urinary
infections
in spinal
patients
125
and often polymicrobial communities that colonize the catheterized urinary tract. It is also obvious from Table II that strains of E. coli possessing the full battery of virulence markers are rarely isolated from the urine of spine-injured patients. There was a higher incidence of virulence factors in the 0-serotypes generally associated with urinary tract infections, than in the autoagglutinable strains and the non-urinary 0-serotypes, but overall the occurrence of these factors was much lower than in collections of E. coli from cases of pyelonephritis in non-catheterized patients (Table III). Only 17% of the strains from the spinal patients produced P-fimbriae, an incidence similar to that reported for normal faecal isolates (18%) and in marked contrast to the 80% recorded for pyelonephritic E. coli.23,25*26It is interesting that while 21 of the strains were capable of expressing MRHA, nine of these did not produce P-fimbriae, suggesting that these particular strains were producing other types of haemagglutinins, the X adhesins28 which have different receptor specificity and tissue tropism from the P-fimbriae.29 The results with aerobactin and haemolysin (Table III) suggest that these systems which provide iron for the growth of pyelonephritic strains’6,30 are not necessary for the stable colonization of the catheterized tract. Mobley et ~2.~’ reported that most E. coli isolated from the urine of catheterized geriatric patients expressed MSHA. The proportions of isolates from short (1 week), medium (2-11 weeks) and long (> 12 weeks) episodes of bacteriuria which expressed MSHA was reported as 59, 65 and 92% respectively. These authors therefore proposed that the ability of E. coli to persist in the catheterized urinary trrlct may depend on the expression of the type 1 fimbriae responsible for MSHA and that other virulence factors may not be required for persistent colonization and pathogenesis in these circumstances. Our study provides some support for this hypothesis in that 76% of our isolates produced MSHA (Table III). Analysis of the 28 defined episodes of E. coli bacteriuria showed that the six most stable colonizers that produced episodes which lasted for 4 weeks or more, were all MS+. In addition it was observed that MS+MRstrains were capable of colonizing the catheterized tract for up to 35 days. Lomberg et a1.21 noted that in young girls with vesicoureteral reflux, pyelonephritic infections were often associated with organisms other than E. cob and that infecting strains of this species rarely carried the virulence factors that were found in the strains producing pyelonephritis in girls who did not have reflux. Vesicoureteral reflux can be common in some spine-injured patients.3’ The data presented in Table IV shows that E. coli was present in only nine of the 22 incidents in which pyelonephritis was diagnosed on the basis of clinical symptoms (fever, nausea, rigors) during the study. It is also clear that these E. coli were poorly endowed with virulence factors (Table V). In conclusion it seems that virulence factors indicative of pyelonephritic strains cannot be used to predict that an E. coli infection in these patients
126
J. Benton
et al.
will lead to an episode of symptomatic pyelonephritis. The neuropathic bladder and the catheterization process allows a variety of organisms that are not virulent for the normal urinary tract to persist in the bladder urine and initiate pyelonephritis. It thus seems that in this compromised group of patients, investigation of host rather than bacterial markers might be a more fruitful approach to the early detection of those cases likely to progress to pyelonephritis. References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.
12.
13.
14.
Newman E, Price M. Bacteriuria in patients with spinal cord lesions: its relationship to urinary drainage appliances. Arch Phys Med Rehabil 1977; 58: 427430. Donovan WH, Stolov WC, Clowers DE, Clowers MR. Bacteriuria during intermittent catheterization following spinal cord injury. Arch Phys Med Rehabill978; 59: 35 l-3.57. Erikson RP, Merritt JL, Opitz JL, Istrup DM. Bacteriuria during follow-up in patients with spinal cord injury. Arch Phys Med Rehabil 1982; 63: 409412. Stickler DJ, Wilmot CB, O’Flynn JD. The mode of development of urinary infection in intermittently catheterized male paraplegics. Paraplegia 1971; 8: 243-252. Clayton CL, Chawla JC, Stickler DJ. Some observations on urinary tract infections in patients undergoing long-term bladder catheterization. J Hasp Infect 1982; 3: 3947. Fawcett C, Chawla JC, Quoraishi A, Stickler DJ. A study of the skin flora of spinal cord injured patients. J Hosp Infect 1986; 8: 149-158. Slade N, Gillespie WA. The Urinary Tract and the Catheter. New York: John Wiley 1985; 49. Warren JW, Muncie HL, Bergquist EJ, Hoopes JM. Sequelae and management of urinary infection in the patient requiring chronic catheterization.? Ural 1981; 125: l-8. Stickler DJ, Chawla JC. The role of antiseptics in the management of patients with long-term indwelling bladder catheters. J Hasp Infect 1987; 10: 219-228. Galloway A, Green HT, Windsor JJ, Menon KK, Gardner BD. Serial concentrations of C-reactive protein as an indicator of urinary tract infection in patients with spinal injury. J Clin Path 1986; 39: 851-855. Stickler DJ, Chawla JC. An appraisal of antibiotic policies for urinary tract infections in patients with spinal cord injuries undergoing long-term intermittent catheterization. Paraplegia 1988; 26: 215-225. Hagberg L, Jodal U, Korhonen TK, Lidin-Janson G, Lindberg U, Svanborg-Eden C. Adhesion, haemagglutination and virulence of Escherichia coli causing urinary tract infections. Infect Immun 1981; 31: 564-570. Winberg J. Balance between host defence and bacterial virulence in urinary tract infection. In: Asscher AW, Brumfitt W, Eds. Microbial Diseases in Nephrology. New York: John Wiley 1986; 197-223. O’Hanley P, Low D, Romero I, Lark D, Vosti K, Falkow S, Schoolnik G. Gal-Gal binding and haemolysin phenotypes associated with uropathogenic Escherichia coli. N
EnglJ Med 1985; 313: 414-420. 15. Johnson JR. Virulence factors in Escherichia coli urinary tract infection. Clin Microbial Revs 1991; 4: 86128. NH, Williams PH. Detection of synthesis of the hydroxamate siderophore 16. Carbonetti aerobactin by pathogenic isolates of Escherichia coli. In: Sussman M, Ed. The Virulence of the Society for General of Escherichia coli: Reviews and Methods. Special publications Microbiology 13. London: Academic Press 1985; 419424. NH, Boonchai S, Parry SH, Vaisanen-Rhen V, Korhonen TK, Williams 17. Carbonetti PH. Aerobactin-mediated iron uptake by Escherichia coli isolates from human extra-intestinal infections. Infect Immun 1986; 51: 966-968. and non-fimbrial haemagglutinins of 18. Duguid JP, Clegg S, Wilson MI. The fimbrial Escherichia coli. J Med Microbial 1979; 12: 213-227. U, Hanson A, Jodal V, Lidin-Janson G, Lincoln K, Olling S. Asymptomatic 19. Lindberg bacteriuria in schoolgirls: differences in Escherichia coli causing asymptomatic and symptomatic bacteriuria. Acta Paediatr Band 1975; 64: 432436.
Urinary
infections
in spinal
patients
127
20. Mobley HLT, Chippendale GR, Tenney JH, Hull RA, Warren JW. Expression of type 1 fimbriae may be required for persistence of Escherichia coli in the catheterized urinary tract. 7 Clin Microbial 1987: 25: 2253-2257. 21. Lomb&g H, Hellstrom H, Jbdal V, Leffler H, Lincoln K, Svanborg-Eden C. Virulence associated traits in Escherichia coli causing first and recurrent episodes of urinary tract infection in children with and without vesicoureteral reflux. J Infect Dis 1984; 150: 561-569. 22. Lidin-Janson G, Hanson LA, Kaijser B et al. Comparison of Escherichia coli from bacteriuric patients with those from faeces of healthy schoolchildren. J Infect Dis 1977;
136: 346-353. 23. Kallenius G, Jacobson SH, Tullus K, Svenson SB. P-fimbriae studies on the diagnosis and prevention of acute pyelonephritis. Infection 1985; 13: 159-161. 24. Funfstuck R, Tschape H, Stein G, Kunath H, Bergner M, Wessel G. Virulence properties of Escherichia coli strains in patients with chronic pyelonephritis. Infection 1986; 14: 145-150. 25. Enerback S, Larsson AC, Leffler H et al. Binding to galactose a l-4 galactose containing receptors a potential tool in urinary tract infection. J Clin Microbial 1987; 25: 407-411. 26. Westerlund B, Sitonen A, Elo J, Williams PH, Korhonen TK, Makela PH. Properties of Escherichia coli isolates from urinary tract infections in boys. J Infect Dis 1988; 158: 996-1002. 27. Warren JW, Tenney JH, Hoopes JM, Muncie HL, Anthoney WC. A prospective microbiologic study of bacteriuria in patients with chronic indwelling uretheral catheters. r Infect Dis 1982; 146: 719-723. 28. Vaisanen-Rhen V. Fimbria-like haemagglutinin of Escherichia coli 075 strains. Infect Immun 1984; 46: 401-407. 29. Nowicki B, Truong L, Movids J, Hull R. Presence of the Dr receptor in normal human tissues and its possible role in the pathogenesis of ascending urinary tract infection. AmJ Path 1988; 133: 14. 30. Calveri SJ, Bohach G, Snyder IS. Escherichia coli o-haemolysin, characteristics and probable role in pathogenicity. Microbial Revs 1984; 48: 326-343. 31. Lamid S. Long-term follow-up of spinal cord injury patients with vesicoureteral reflux. Paraplegia 1988; 26: 27-34.
