Clustered multidrug-resistant Bordetella petrii in adult cystic fibrosis patients in Ireland : case report and review of antimicrobial therapies

Received 23 September 2013 Accepted 30 January 2014 Graduate Entry Medical School and Centre for Interventions in Infection, Inflammation & Immunity (4i), University of Limerick, Limerick, Ireland University Hospital Limerick, Dooradoyle, Limerick, Ireland Tallaght Hospital, CFAI Reference Laboratory, Dublin, Ireland Trinity College Dublin, Clinical Microbiology Department, Dublin, Ireland AMRHAI Reference Unit, Reference Microbiology Services, Public Health England, London, United Kingdom

Globally, Ireland has the highest prevalence of CF (a recessive autosomal disease associated with increased susceptibility to respiratory infection) with a prevalence of 2.98 per 10 000 of the population compared with prevalence in Europe and the USA of 0.737 and 0.797 per 10 000, respectively (Farrell, 2008).Bordetella species have Abbreviations: bd, twice day 21 ; i.v., intravenously; MRSA, meticillinresistant Staphylococcus aureus; MSSA, meticillin-sensitive S. aureus; od, once day 21 ; p.o., by mouth; tds, three times day 21 .
been associated with CF, with B. petrii in particular being the species commonly cultured in the sputum of these patients (Spilker et al., 2008).However, the clinical significance of B. petrii in CF is unknown.In this report, using a case series of adult CF patients colonized by a single discrete strain of B. petrii and with varying baselines of health (one of whom was asymptomatic), we aimed to correlate colonization with incidence of acute exacerbation of symptoms.In addition, as definitive guidelines for antimicrobial sensitivity/resistance do not exist for B. petrii, we completed a systematic review of the available literature to collate evidence of antimicrobial efficacy against B. petrii and compared this information with the characteristics of the isolate obtained from the four patients in this study with a view to determining effective antimicrobial therapy.

Case report
Four CF patients (three female, mean age 21 years, nonsmokers), named A-D, with chronic B. petrii colonization were identified in University Hospital, Limerick, Ireland, between March 2009 and December 2012.Retrospective analyses of patient charts and stored sputum samples determined the frequency of hospitalization, episodes of infections, antimicrobial use and patient quality of life scores (Henry et al., 2003) for the year prior to B. petrii detection and throughout colonization.Emphasis was placed on the presence/absence of clinical features during colonization, in particular lung function, efficacy of antimicrobial therapy and patient outcome.
Genotypically, patients A and B were heterozygotic for the Df508 mutation (within the CF transmembrane conductance regulator causing loss of the amino acid phenylalanine, affecting chloride ion channels in cell membranes and leading to production of thickened mucus), whilst patients C and D were homozygous.All resided in the same county in the midwest of Ireland.Each of the patients had experienced at least one acute hospitalization in the year prior to B. petrii detection.As they attended the same outpatient clinic, exposure to one another was possible.The clinic implements a system of segregated clinics according to pathogen status [in particular for Burkholderia cepacia complex, Pseudomonas aeruginosa and meticillin-resistant Staphylococcus aureus (MRSA)].Each clinic uses singlepatient rooms, with healthcare workers rotating through these (i.e.patients remain in situ to limit cross-transmission of micro-organisms through contact with other patients, and the healthcare workers adopt strict infection control practices during patient contact).However, in 2003, patients A and D were admitted simultaneously to the same paediatric ward for 1 day.Also in 2008, patients A and C were hospitalized simultaneously but in separate wards.B. petrii was initially detected in March 2009.
B. petrii from sputum was preliminarily reported as an undetermined Bordetella species.Subsequent identification of isolates from the four patients was by PCR 16S rRNA gene sequencing and gyrB sequence cluster analysis, PFGE and matrix-assisted laser desorption/ionization-time of flight techniques, and indicated the presence of a single discrete B. petrii strain.Our group has found that PFGE is capable of differentiating among B. petrii isolates (data not shown).Currently, there is no definitive classification of B. petrii as sensitive/resistant to specific antimicrobials.In this study, the efficacy of selected antimicrobials against the B. petrii isolates (subsequently found to be a single strain) was assessed using Etest (bioMe ´rieux) strips.1).Monomicrobial culture of B. petrii was relatively infrequent, with P. aeruginosa being the most common co-isolate, although meticillin-sensitive S. aureus (MSSA) was found occasionally in three of the four patients, as were Candida species (Table 1) In patient A, B. petrii was first co-isolated with Aspergillus fumigatus.Ten days later, the patient required antimicrobial therapy at home [piperacillin/tazobactam, 4.5 g three times day 21 (tds) intravenously (i.v.)], nebulized colomycin [2 million units twice day 21 (bd)] and clarithromycin [500 mg bd by mouth (p.o.)].Subsequently, B. petrii was not detected for 10 months until February 2010 when B. petrii re-occurred as a monomicrobial isolate.Patient A was treated with nebulized colomycin (as above) and a 3 week course of oral ciprofloxacin (500 mg bd).In April 2010, treatment for P. aeruginosa was nebulized colomycin (as above) and meropenem (2 g tds i.v.), and in May 2010, B. petrii was detected for the third time.Aspergillus fumigatus was also present.At that time, the patient presented with a 3-day history of haemoptysis and increased volume of sputum.A chest X-ray showed patchy infiltrates.Successful treatment consisted of 2 weeks of nebulized colomycin (as above) and meropenem (as above) and switch therapy with oral co-amoxiclav (625 mg tds).In August 2010 and April 2011, B. petrii was again detected, although the patient was asymptomatic on both occasions.B. petrii was not detected again until December 2011 but now persisted concomitantly with Achromobacter xylosoxidans and Aspergillus fumigatus.
In patient B, following first detection, B. petrii was identified in 12 of the subsequent 13 sputum samples tested (92 %).Nine of these were polymicrobial, and only one with P. aeruginosa.Other co-colonizers included Candida spp., MSSA, Cupriavidus metallidurans, Alcaligenes faecalis and Achromobacter xylosoxidans.No respiratory symptoms were reported and therefore no antimicrobial treatment was offered.
In contrast, patient C was positive for B. petrii on only 3 of 13 samplings (23 %) since first detection.On each occasion, P. aeruginosa was also identified, and the patient required treatment for infective exacerbation of CF.In January 2011, treatment comprised tobramycin [10 mg kg 21 once day 21 (od) i.v.] (dependent on trough levels), flucloxacillin [2 g four times day 21 (qds) i.v.] and piperacillin/tazobactam (4.5 g tds i.v.), which effectively eradicated B. petrii until June 2011, at which time P. aeruginosa was also cultured.The patient was admitted for 4 days of i.v.hydrocortisone treatment with a course of ciprofloxacin (500 mg bd p.o.).However, the patient clinically deteriorated after cessation of treatment and merited further antimicrobials for 2 weeks in July: a repeat course of ciprofloxacin (as above) with linezolid (600 mg bd p.o.).In August 2011, patient C was again admitted for acute infective exacerbation coinciding with B. petrii, P. aeruginosa and MRSA being isolated from the sputum.Symptoms included a history of 6 kg weight loss over the prior 5 weeks, 2 weeks of shortness of breath at rest, a productive cough with yellow sputum and pyrexia.
Patient D was admitted due to acute infective exacerbations of CF in May, September and November 2011 coinciding with detection of monomicrobial B. petrii.Antimicrobial treatment in May comprised meropenem (2 g tds i.v.), tobramycin (10 mg kg 21 od i.v.) (dependent on trough levels) and linezolid (600 mg bd p.o.); the latter was cycled with flucloxacillin (2 g qds i.v.) plus meropenem (as above) and tobramycin (as above) for the subsequent exacerbation in September.During the hospitalization in December 2011, patient D developed a right pneumothorax (.50 %).Eight days later, additional antimicrobials were administered, specifically, vancomycin (15 mg kg 21 bd i.v.), piperacillin/tazobactam (4.5 g tds i.v.) and azithromycin (500 mg od p.o.) with ciprofloxacin (500 mg bd p.o.) to treat B. petrii, P. aeruginosa and Candida sp.These proved ineffective, and 4 days later, patient D died due to cardiorespiratory failure.In performing analyses for this study, we determined that, from the adult CF patient population of 45 in Limerick, B. petrii had colonized four patients (9 %).In contrast, none of the 81 paediatric CF patients have been positive for B. petrii, supporting the rationale that B. petrii, as an environmental isolate, is emerging as an opportunistic pathogen in patients with altered microflora due to repeated, long-term exposure to antimicrobials (Zelazny et al., 2013;Gross et al., 2008;von Wintzingerode et al., 2002).However, the concept of emergent pathogens should be considered in the context of better diagnostic techniques and their influence on reported incidences.
In the four patients described here, B. petrii was monomicrobial in only 16 % of positive samples, and was found in conjunction with known pathogens in 68 % of microbiological processing of sputa (e.g.P. aeruginosa was detected in 30 %) and with non-pathogens (Alcaligenes faecalis, Aspergillus fumigatus and C. metallidurans) in 10 %.Owing to the complex nature of CF infections with the tendency of bacteria to co-exist as biofilms within the lungs of CF patients, as well as varying  (Fry et al., 2005).However, in vivo antimicrobial treatments did not mirror results under laboratory conditions and, specifically, the VITEK2 Compact system has been criticized as inappropriate for the susceptibility testing of Bordetella species (Le Coustumier et al., 2011).In practice, clarithromycin has been used preferentially for B. petrii.However, studies have shown resistance to this antibiotic in vitro, and when administered to patients for differing durations -6 weeks (Fry et al., 2005) and 8 weeks (Stark et al., 2007) -improvement of symptoms was observed but in conjunction with the emergence of resistance.Indeed, positive patient outcome was subsequently attributed to ancillary treatment (Fry et al., 2005).Amoxicillin and metronidazole have been administered with no resolution of symptoms (Fry et al., 2005), whilst co-amoxiclav has improved symptoms but not eradicated B. petrii colonization (Le Coustumier et al., 2011).In the patients reported here, B. petrii was sensitive to piperacillin/tazobactam and minocycline but resistant to erythromycin and aztreonam (Table 2).Worryingly, the Limerick B. petrii isolate was resistant to antimicrobials in the macrolide, most of the b- lactam and the fluoroquinolone groups, and can therefore be considered multidrug-resistant.
Overall, this paper advances our knowledge surrounding B. petrii in the clinical setting, and represents the first report to our knowledge of multiple patients colonized by a single B. petrii strain, highlights problematic B. petrii identification and indicates that colonization may be underestimated in the CF population.Furthermore, B. petrii    (Fry et al., 2005;Le Coustumier et al., 2011;Stark et al., 2007;von Wintzingerode et al., 2001) and the four patients in this study.
was present in all cases of CF exacerbation in the patients colonized, and was found to be multidrug-resistant, with piperacillin/tazobactam being the only effective common antimicrobial therapy.
Patients A, B, C and D first presented with B. petrii in March 2009, March 2010, January 2011 and February 2011, respectively.Durations of intermittent colonization ranged from 8 to 35 months (Table . Notably, three of the four patients (A, C and D) experienced at least one acute infective exacerbation of CF while simultaneously being positive for B. petrii.These patients were treated with antimicrobials directly following detection of B. petrii.Patient B was neither admitted to hospital nor treated for infective exacerbation during the course of B. petrii colonization.Retrospective analysis of spirometry results for all four patients failed to indicate a correlation between B. petrii detection and impaired lung function [measured as forced expiratory volume in 1 s (FEV1)].

Table 1 .
Summary of information from patients A-D of the case series Stark et al., 2007]udies in which varying media were used to culture B. petrii [fastidious anaerobe agar, Columbia blood agar and MacConkey agar, horse blood, and MacConkey and chocolate agar, as well as chocolate PolyViteX, bromocresol purple and selective Haemophilus agar (chocolate bacitracin);Stark et al., 2007], we found that conventional microbiology techniques for identification of B. petrii were unsuccessful and, indeed, often misidentified the bacterium.However, molecular techniques were effective.
baselines of patient health, it is difficult to definitively state the clinical significance of B. petrii in this case series.However, two of the patients were symptomatic during monomicrobial colonization by B. petrii, and one patient was symptomatic during co-colonization by B. petrii and Aspergillus fumigatus, a species with limited evidence of invasive pathogenicity in CF.Conversely, however, patient B experienced persistent asymptomatic colonization for 22 months, and the presence of B. petrii could not be correlated with variations in lung function for any of the four patients.The cases reported here illustrate the recalcitrance of B. petrii and its ability to recur.Previously published papers, describing clinical involvement of B. petrii, have failed to identify the sources of infection, and no guidelines exist as yet regarding the antimicrobial susceptibility of B. petrii.Antimicrobial MICs in the literature are the only guidance for effective treatment of B. petrii, and so this study brings together for the first time (Table2) MICs for the first B. petrii environmental isolate and available information for all published clinical isolates and the multiple isolates of the Limerick strain from this study (completed following Clinical and Laboratory Standards Institute guidelines).In summary, susceptibility testing using disc diffusion tests showed apparent sensitivity of B. petrii to erythromycin, gentamicin, ceftriaxone and piperacillin/tazobactam, with resistance to amoxicillin, co-amoxiclav, tetracycline, clindamycin, ciprofloxacin and metronidazole

Table 2 .
Summary of MIC values (mg l 21