Prune belly syndrome

Abstract: Klebsiella pneumoniae resistant to third-generation cephalosporins and gentamicin was isolated from two patients in a paediatric intensive care unit within a two-week period. The double-disc diffusion test indicated the presence of an extended-spectrum (beta)-lactamase (ESBL). The unit was closed to admissions, and stringent infection control procedures were implemented. Environmental screening and screening of staff and patients on the unit were commenced. Two weeks later, K. pneumoniae with an identical antibiogram was isolated from the urine of a patient in a different ward. Blood-culture isolates possessed the K16 antigen, while the urine isolate was non-typeable. The isolates were shown to be similar when banding patterns of XbaI chromosomal DNA digests were compared. The resistance to the extended-spectrum cephalosporins was shown to be transferable in association with a large plasmid >98 mDa. Resistance to gentamicin always co-transferred with 5lactamase resistance and appeared to be encoded by the same plasmid.

Key words: beta-Lactamases. Klebsiella pneumoniae.

Introduction

The incidence of infection with resistant Gramnegative rods has increased worldwide.1,2 Of particular concern is the emergence of extended-spectrum (beta)-lactamase (ESBL)-producing Klebsiella pneumoniae, which has been a serious cause of nosocomial infection throughout the United States and Europe.3 Emergence of this resistance has, in part, been attributed to the growing use of extended-spectrum cephalosporins, the introduction of which in the early 1980s shortly preceded the initial descriptions of ESBL-producing Enterobacteriaceae in 1987.4 Recognised risk factors of infection with resistant organisms include treatment in intensive care units, recent surgery, prolonged hospital stay, and antibiotic exposure.5 The first cases of infection with ESBL-producing At. pneumoniae in this hospital, involving three patients, were identified in May 1995. To the best of our knowledge, this is the first reported occurrence of such in the Republic of Ireland.

Case histories

Patient 1

On 1 April 1995, a two-year-old male child with Prune Belly syndrome, who had been maintained on home peritoneal dialysis, was hospitalised with suspected peritonitis. He had received several course of antibiotics, often with vancomycin and/or extendedspectrum cephalosporins. He was admitted to a general paediatric ward, and treated with empiric intraperitoneal vancomycin and ceftazidime, and intravenous cefotaxime. His condition improved steadily over the first 10 days. Thereafter, however, relapsing peritonitis prompted further treatment, first with ceftazidime and gentamicin, and subsequently with vancomycin, ticarcillin and gentamicin. Peritoneal dialysis was discontinued and the peritoneal catheter removed. He was transferred to a single room in the paediatric intensive care unit (ICU) and continuous veno-venous haemodiafiltration commenced. Throughout this time, the white cell count (WCC) in the peritoneal fluid (PF) fluctuated between 93 and >2000 cells/mm^sub 3^, and repeated cultures of PF and blood were negative.

On 27 April, ulceration of the buttocks and perineum was noted. Although receiving vancomycin, ticarcillin, gentamicin and fluconazole, the latter added because of candiduria, blood obtained for culture just prior to his transfer to the ICU yielded K. pneumoniae. K. pneumoniae was also isolated from a buttock swab. Bacteraemia persisted throughout the next day, and antibiotic susceptibility testing indicated resistance to gentamicin, ticarcillin and ceftazidime. Imipenem and amikacin were commenced, with dose adjustment for renal failure. Blood samples obtained for culture on 1 May, 4 May and 8 May were sterile.

His clinical condition deteriorated with increasing abdominal distention, progressive peritonitis, and onset of melaena. Culture of the peritoneal fluid now yielded Candida albicans and liposomal amphoteracin B (3 mg/kg) was commenced. Despite continued therapy with imipenem and amikacin, blood obtained on 9 May yielded K. pneumoniae. Bacteraemia persisted for a further four days despite adequate amikacin levels (peak: 30.15 mg/L, trough: 4.5 mg/L), and resolved coincident with the addition of ciprofloxacin to the therapeutic regimen and removal of the central venous catheter (CVC). K. pneumoniae was also recovered from the CVC line tip. Thereafter, blood cultures remained sterile, and antibacterial agents were discontinued after a further 10 days of therapy. Candida peritonitis persisted, refractory to intravenous and intraperitoneal amphoteracin and 5-flucytosine, and was the ultimate cause of death.

Patient 2

The single room adjacent to that of the index patient was occupied by a five-month-old male infant, admitted to the hospital as a newborn, with a congenital diaphragmatic hernia. He also suffered from severe exfoliative dermatitis and had chronic lung disease. He had received numerous courses of broad-spectrum antimicrobial agents, including anti-pseudomonal penicillins, extended-spectrum cephalosporins, aminoglycosides and glycopeptides. On 9 May 1995, while completing a course of teicoplanin, ceftazidime and amikacin for Pseudomonas aeruginosa pneumonia and catheter-related infection with a coagulasenegative Staphylococcus, he became pyrexial (t = 39(deg)C) and appeared clinically septic. His peripheral WBC count rose to 55.6 x 10^sub 9^/L with 86% neutrophils. Blood culture yielded Klebsiella pneumoniae the following day. Ceftazidime was discontinued and imipenem commenced. Clearance of bacteraemia was documented on 12 May, and he completed 10 further days of imipenem and amikacin therapy.

Patient 3

On 25 May 1995 a five-month-old female infant, who had surgical repair of a meningomyelocoele after birth and a ventriculo-peritoneal shunt inserted at one month of age, was admitted with a urinary tract infection (UTI). She had had one previous UTI with Escherichia coli at four months of age, after which she was maintained on amoxycillin-clavulinate as urinary prophylaxis. On this occasion, resistant K. pneumoniae with an identical antibiogram to that of the isolates from patients 1 and 2 was isolated from the urine. On review of her chart, it was noted that three months earlier, in February 1995, K. pneumoniae with an identical antibiogram had been isolated from a swab taken from the repaired meningomyelocoele site. In effect, this was the first patient to become colonised with this organism. She was never admitted to the ICU, nor was there any obvious overlap in medical carers between this patient and the other two. However, the possibility of such overlap can not be excluded.

Because this was the first recognised incidence of ceftazidime-resistant K. pneumoniae infection at this hospital, we sought to define the problem in terms of spread of the organism within the ICU, and to further characterise the nature of the ceftazidime resistance.

Laboratory materials and methods

Cultures

Blood was cultured using BACTEC PLUS Anaerobic/F, and BACTEC Paeds PLUS/F bottles in a BACTEC 9240 (Becton Dickinson). Direct Gram stains were examined on all positive blood cultures. Positive cultures were sub-cultured onto Columbia agar containing 7% horse blood, both aerobically and anaerobically, Columbia chocolate agar in 7% CO2, and aerobically on MacConkey agar without salt. Where indicated, quantitative urine cultures were carried out on MacConkey agar without salt, using a 0.1 mL standard loop. Buttock, skin, and drain-site swabs were cultured onto Columbia agar containing 7% horse blood in 7% CO2, Columbia chocolate agar in 7% CO2, and aerobically on MacConkey agar without salt. Environmental screening swabs were cultured onto Columbia agar containing 7% horse blood, and MacConkey agar without salt. All culture media were supplied by Oxoid.

Biotyping

Gram-negative bacilli were identified using the API 20E system (API-bioMerieux) according to manufacturer's instructions.

Antibiograms

Susceptibilities were determined by disc diffusion on Oxoid Isosensitest agar. Zone sizes were compared with E. coli NCTC10418 and E. coli NCTC11562 for (beta)-lactam-inhibitor combinations as recommended by the British Society for Antimicrobial Chemotherapy.6 The double-disc synergy test was used to detect the presence of an ESBL by placing a 30 Rg augmentin disc 20 mm from a 30 (mu)g ceftazidime disc on an inoculated Isosensitest agar plate.1 At a later date when ESBL Etest (Cambridge Diagnostic Services Ltd) became available, retrospective examination of the isolates was carried out according to the manufacturer's instructions. Etest minimum inhibitory concentrations (MIC) to tazobactam-piperacillin combinations were also carried out.

Epidemiological typing

All isolates of K. pneumoniae were serotyped by counter-current immunoelectrophoresis of capsular antigens.7 To clarify the epidemiology of these isolates, banding.patterns of XbaI chromosomal digests were compared by computer analysis following pulsed-field gel electrophoresis.

Plasmid analysis and conjugation studies

Transfer of resistance to ceftazidime and gentamicin was investigated using a cross-streak mating method with Klebsiella oxytoca strain 478 (Rif ) or E. coli strain J62-2 (Rif) as recipients. Following incubation of donor and recipient mixtures overnight at 37(deg)C, transconjugants were selected on nutrient agar containing rifampicin (50 mg/L) and either ceftazidime (4 mg/L) or gentamicin (10 mg/L). Transfer of aminoglycoside resistance as well as cephalosporin resistance was investigated, as genes encoding aminoglycoside resistance are often linked with genes encoding ESBLs.

Plasmid DNA was extracted from the clinical isolates and selected transconjugants.9 The MICs of ceftazidime and amoxycillin (both agents tested with and without clavulanic acid), pipericillin (with and without tazobactam) and gentamicin were determined for representative transconjugants using Etest.

Infection control measures

Following the isolation of the resistant K. pneumoniae from patient 2, the ICU was closed to further admissions for one week, in order to determine the extent of the problem and to prevent further cases of crossinfection. Patients 1 and 2 were cohort nursed in adjacent rooms within the unit. Contact isolation precautions were implemented; gloves and disposable plastic aprons were worn by all staff when direct contact with the patient or his environment was anticipated. Medical equipment was dedicated to single patient use. Housekeeping practices were intensified throughout the unit. Informal education programmes relating to the mode and prevention of spread of K. pneumoniae were held. Particular emphasis was placed on the role of hand-washing in the prevention of transmission. Access to the unit was restricted to essential personnel only.

Patient and environmental screening

Surveillance screening of all patients in the unit was carried out. Stools, urine, skin and drain-site swabs were taken for culture from all patients. Patients were discharged from the unit only if surveillance cultures were negative and their medical condition permitted. Environmental screening included horizontal surfaces and patient care equipment located in the patient rooms, sluice room, treatment room and storage areas. Limited environmental screening was performed in the general paediatric ward to which patient 1 had been first admitted. The hands of all ICU staff and residents were examined for cuts and lesions, and hands were screened for possible carriage of Klebsiella. As patients 1 and 2 had been fed by naso-gastric tube prior to commencement of total parenteral nutrition (TPN), the hospital formula room was surveyed, and feed lists were cross-referenced.

Laboratory results

Cultures

K. pneumoniae was isolated from the blood culture and buttock swabs of patient 1, from the blood cultures and skin swabs of patient 2, and from the urine of patient 3. Susceptibility test results were identical for all three isolates and are shown in Table 1. A zone of inhibition, indicative of the presence of an ESBL, can be seen between the augmentin and ceftazidime discs (Fig. 1). Isolates from patients 1 and 2 were of the K16 serotype, and that of patient 3 was non-typeable.

Etest and ESBL Etest(R) results

MIC of ceftazidime was >32 (mu)g/mL. MIC of ceftazidime with clavulanic acid ranged from 8 to 32 (mu)g/mL and was inoculum-dependent. MIC of tazobactam was >32 (mu)g/mL.

Molecular typing

Strain similarity was confirmed by molecular typing which showed similar banding patterns of XbaI chromosomal digests (Fig. 2). Plasimd profiles

PLasmid profiles were carried out on blood culture isolates only. Isolates from patients 1 and 2 showed identical profiles (data not shown).

Transfer experiments

Resistance to ceftazidime was transferred to recipient strains of K. pneumoniae and E. coli in association with a large plasmid >98 mDa. The MICs shown in Table 2 are those of the recipients and two transconjugants, one derived from each recipient. Changes in MIC of the transconjugants therefore indicated resistance associated with acquisition of the >98 mDa plasmid. These data confirmed that the clinical isolates contained a transferable ESBL which conferred resistance to ceftazidime and other (beta)-lactams, but was inhibited by clavulanic acid and tazobactam. Resistance to gentamicin always co-transferred with (beta)-lactam resistance. It appeared to be encoded by the same plasmid because direct selection for the transfer of gentamicin resistance also yielded transconjugants with the >98 mDa plasmid which were also resistant to ceftazidime.

Patient and environmental screening

Screening failed to detect colonisation with resistant K. pneumoniae in any other of the six patients in the ICU at the time. Three long-stay patients who were on the same ward as patient 1 before transfer to ICU were also screened and were negative. Resistant K. pneumoniae was not isolated from the hands of any of 64 medical staff. It must be noted, however, that stringent infection control measures had been implemented and the importance of hand-washing techniques was emphasised during the screening period. K. pneumoniae was isolated from an armchair in the room occupied by patient 2. All other 177 environmental samples were negative. Swabs taken from 61 sites in the general paediatric ward were negative. Screening failed to define an epidemiological link between cases in the ICU and case 3. However, the possibility of overlap in medical or surgical carers between these patients can not be excluded.

Discussion

Risk factors for emergence of infection with resistant Gram-negative rods include treatment in ICUs, recent surgery, prolonged hospital stay, instrumentation, and antibiotic exposure.5 Each of the patients in this study had at least three of these risk factors.

Direct patient-to-patient transfer seems likely between patients 1 and 2 as they were nursed in proximity in an ICU, the time courses of infection overlapped and strain similarity was proved. The possibility that each patient acquired infection from a third, unidentified source is unlikely as the index patient first had the microorganism isolated while on a general paediatric ward with which patient 2 had no contact. Further, recovery of the isolate from the armchair in the room occupied by patient 2 demonstrated some limited environmental contamination within the unit. Although K. pneumoniae was not recovered from the hands of any staff member, screening took place after implementation of infection control measures, and a campaign to heighten awareness of the importance of hand-washing in curtailing spread within the unit.

While patient 3 had no obvious contact with the other patients, the organism isolated from her urine showed identical banding patterns of chromosomal DNA digests, and was considered to be the same. Review of case notes confirmed that she was the first patient to acquire this organism, two months before infection was documented in patient 1.

The isolates were initially considered to be the same or similar, due to their unusually resistant antibiograms. They were resistant to third-generation cephalosporins, and to netilamicin and gentamicin.

An unusual and consistent feature was a zone of inhibition between the ceftazidime and augmentin discs. The zone indicated the presence of an ESBL which could be inhibited by clavulanic acid. There was no zone around the augmentin disc. It was thought probable that the K. pneumoniae isolates produced large quantities of (beta)-lactamase, and that inhibition of clavulanic acid was insufficient to restore amoxycillin sensitivity. This was confirmed by MICs of transconjugants, which showed that although inhibition of the ESBL by clavulanic acid resulted in a 10- to 20-fold reduction in the MIC of augmentin compared to amoxycillin, this decrease was not sufficient to restore sensitivity. Sensitivity to ceftazidime was, however, fully restored by the addition of clavulanic acid. Ceftazidime was probably easier to protect than amoxycillin. The level of (beta)-lactam inhibitor required to potentiate a (beta)-lactam drug varies with the drug. Factors involved include the permeability of the drug, the relative affinities of the drug for the penicillin-binding protein and the (beta)-lactamase, and the efficiency of the enzyme.10,11 MICs of transconjugants showed that the ESBL was inhibited by tazobactam, the MIC was 2 (mu)g/mL compared to 256 (mu)g/ mL for piperacillin alone. However, there was no detectable zone on disc sensitivity testing of the clinical isolates, and the MIC determined by Etest for piperacillin:tazobactam was >256 (mu)g/mL. Similarly, MICs for augmentin and clavulanic acid/ceftazidime combinations were much higher than those of the transconjugants. The clinical isolates may have produced an increased amount of SHV chromosomal (beta)-lactamase, or possessed a TEM-1 enzyme. In this case, the tazobactam or clavulanic acid would have been overwhelmed by the (beta)-lactamase produced by the isolate. Alternatively, the clinical isolates may have been less permeable than the recipient E. coli or K. pneumoniae.

Most ESBL enzymes that are inhibited by clavulanic acid, and are not active against cefoxitin, are derivatives of the TEM-1 and SHV-1 enzymes.' They result from mutations which give rise to amino acid substitutions at the active site of the enzyme. Substitutions which broaden the spectrum of activity reduce the efficiency of the enzyme.12 The genes encoding the enzyme tend to have more efficient promoters. This may explain the large amount of (beta)lactamase produced by the isolates.

In the transconjugants, ceftazidime resistance was associated with a large plasmid >98 mDa. Resistance to gentamicin always co-transferred with [-lactam resistance, therefore, it appeared to be coded by the same plasmid. ESBLs are frequently encoded by large, multi-resistance plasmids, and simultaneous resistance to aminoglycosides often occurs. While detection of this (beta)-lactamase was easy using disc sensitivity testing and the double-disc synergy test, it is not always straightforward. Often, an ESBL may increase the MIC, but the breakpoint is not reached. However, failure of treatment may be due to a dramatic rise in MIC with an increased inoculum.13 It may be possible to increase detection by lowering the accepted breakpoint, or by using a 5 jig ceftazidime disc.14 All strains showing aminoglycoside resistance warrant further investigation.

ESBLs are detected most frequently in K. pneumoniae, although the reason for this is unclear; it may be a particularly good vector for gene transfer.12 Although not quantified, our isolates transferred resistance more readily to K. pneumoniae than to E. coli. Most outbreaks of infection with ESBL-producing organisms are K. pneumoniae outbreaks. Crossinfection may be aided by longer survival of K pneumoniae on hands, and greater resistance to washing with soap and water, than other Enterobacteriaceae." Prior antibiotic treatment may encourage intestinal colonisation with resistant Gram-negative rods. Multiple resistance plasmids, such as that harboured by our clinical isolates, have been shown to code for the production of a non-fimbrial adhesion protein, similar to that of enterotoxigenic E. coli. This protein facilitates colonisation of the human intestinal tract.16 In this context, it is interesting that K. pneumoniae was isolated from buttock swabs in one of the patients. We have since encountered colonisation and infection with ESBL-producing K. pneumoniae in the perianal area of an immune-suppressed patient with positive blood cultures.

Investigation of the source of infection failed to show colonisation of hands of staff members, or of the environment. One chair in the ICU was contaminated. Closure of the unit, extensive cleaning, use of barrier precautions, and strict control of the use of ceftazidime was successful in preventing a wider outbreak. In the following 12-month period, 12 other ESBL-producing K. pneumoniae were isolated from patients throughout the hospital. None of these isolates possessed the K16 antigen. Those which failed to type serologically were shown to be distinct from the original strain by pulsed-field gel electrophoresis of chromosomal DNA digests. As the ESBLs were not fully characterised, it was not possible to determine whether the plasmid had been transferred to other K. pneumoniae. However, none of the 12 strains was as resistant to (beta)-lactams as the original isolates, and detection of the enzymes was, at times, difficult. This suggests the involvement of other (beta)-lactamases. Cross-resistance to gentamicin was observed in all cases

There is no suitably reliable test for the detection of ESBL. The double-disc diffusion test is useful, but is dependent upon the distance between the discs, and the amount and efficiency of the enzyme produced. ESBL Etest is too expensive for routine screening but may be used for confirmation. Neither of these tests will detect the emergence of ESBLs resistant to (beta)-lactamase inhibitors.17 A 5 (mu)g ceftazidime disc may help detect both types of enzyme. On detection of an ESBL, the organism should be reported as resistant to all third-generation cephalosporins, even if the pharmaceutical breakpoint is not reached.13 Greater clinical awareness of this problem is essential. There is a need for close monitoring in intensive care and oncology units as ESBL-producing K. pneumoniae can rapidly disseminate within these areas.

We are grateful to the scientific staff of the Gram Negative section at the Laboratory of Hospital Infection, CPHL, Colindale, London, who serotyped the K pneumoniae isolates. We also thank Dr Neil Woodford, Laboratory of Hospital Infection, CPHL, Colindale, London, for the plasmid analysis and conjugation studies, and Ms M. E. Kaufmann, Laboratory of Hospital Infection, CPHL, Colindale, London for epidemiological typing of the strains.

References

Philippon A, Arlet G, Lagrange PH. Origin and impact of plasmid mediated extended spectrum beta-lactamases. Rev Eur J Clin Microbiol Infect Dis 1994; 13(Suppl 1):17-29.

Ehrhardt AF, Sanders CC. Beta-lactam resistance amongst Enterobacter species. I Antimicrob Chemother 1993; 32(Suppl B):l-11.

Jacoby GA, Medeiros AA. More extended spectrum betalactamases. Antimicrob -,Agents Chemother 1991; 35:1697-704.

Bauernfeind A, Horl G. Novel R-factor-borne (-lactamase of Escherichia coli conferring resistance to cephalosporins. Infection 1987; 15:257-9.

Gould IM. Risk factors for acquisition of multiple drugresistant Gram-negative bacteria. Rev Eur J Clin Microbiol Infect Dis 1994; 13(Suppl 1):30-8.

British Society of Antimicrobial Chemotherapy Working Party. A guide to sensitivity testing. J Antimicrob Chemother 1991; 27(Suppl D):1-50.

7 Palfreyman JM. Klebsiella serotyping by counter-current immunoelectrophoresis. J Hygiene 1978; 81:219-25.

8 Kaufman ME, Pitt TL. Pulsed-field gel electrophoresis of bacterial DNA. In: Chart H, ed. Methods in practical laboratory bacteriology. Boca Raton, FL: CRC Press, 1994: Ch 8.

9 Kado CI, Liu ST. Rapid procedure for detection and isolation of large and small plasmids. J Bacteriol 1981; 145:1365-73.

10 Livermore DM. Determinants of the activity of beta-lactamase

inhibitor combinations. J Antimicrob Chemother 1993; 31(Suppl A):9-21.

11 Sanders CC. Beta-lactamases of Gram-negative bacteria. New

challenges for new drugs. Rev Clin Infect Dis 1992; 14:1089-99. 12 Jacoby GA. Genetics of extended spectrum beta-lactamases. Rev Eur* Clin Microbiol Infect Dis 1994; 13(Suppl 1):2-11.

13 Livermore DM. Beta-lactamases in laboratory and clinical resistance. Clin Microbiol Rev 1995; 8:557-84. 14 Jacoby GA, Han P. Detection of extended spectrum P-lactamases in clinical isolates of Klebsiella pneumoniae and Escherichia coli Clin Microbiol 1996; 34:908-11.

15 Casewell MW, Desai N. Survival of multiple resistant Klebsiella aerogenes and other Gram-negative bacteria on fingertips. _ Hosp Infect 1983; 4:350-60.

16 Darfeuille-Michaud A et al. R-plasmid-encoded adhesive factor in Klebsiella pneumoniae strains responsible for human nosocomial infections. Infect Immun 1992; 60:44-55. 17 Pornull KJ, Rodrigo G, Dornbusch K. Production of a plasmid mediated AmpC-like 4-lactamase by a Klebsiella pneumoniae septicaemia isolate. _Antimicrob Chemother 1994; 34:943-54.

JUANITA GROGAN, HELEN MURPHY and KARINA BUTLER

Department of Clinical Microbiology. Our Lady's Hospital for Sick Children, Crumlin Dublin 12, Ireland

(Accepted 27 January 1998)

Copyright Royal Society of Medicine Press Ltd. Jun 1998
Provided by ProQuest Information and Learning Company. All rights Reserved