Epidemic spread of Pandoraea pulmonicola in a cystic fibrosis center

Pandoraea spp. are non-fermentative aerobic Gram negative bacilli first described in 2000 by Coenye et al. [1]. Initially, the genus Pandoraea was comprised of five bacterial species (Pandoraea apista, Pandoraea pulmonicola, Pandoraea pnomenusa, Pandoraea sputorum and Pandoraea norimbergensis). Since then, four other unnamed genomospecies and four other named species (Pandoraea thiooxidans, Pandoraea oxalativorans, Pandoraea faecigallinarum and Pandoraea vervacti) have been characterized [2–4]. These bacteria are ubiquitous in the environment and have been recovered from the soil, sea, and drinking water [5, 6]. In humans, Pandoraea spp. are mainly isolated from Cystic Fibrosis (CF) patients, and may cause chronic lung colonization [7–9]. Several Pandoraea species have been reported in CF patients. P. apista was involved in an epidemic spread in six CF patients in Denmark in 2003 [10]. More recently, P. sputorum was an agent of chronic CF lung colonization in Spain [8]. In addition to lung colonization, invasive infections due to Pandoraea species were also reported, although mostly in non-CF patients [11, 12]. However, a case of bacteraemia due to Pandoraea sp. was described in a CF patient [13]. This study reports an epidemic spread of P. pulmonicola in six patients attending our CF center and is the first description of an epidemic involving this bacterial species.

One of the patients attending our CF center was chronically colonized with P. pulmonicola since 2000, and five out of the 243 patients acquired this bacterium in 2009. Among Pandoraea species, P. pulmonicola was already reported in CF chronic lung colonization. This bacterial species was also reported to be the most predominant Pandoraea species among Irish CF patients [20]. Phenotypical characteristics of strains are consistent with the initial description made by Coenye et al. [1], since all strains displayed catalase and oxidase activity, polar motility and grew aerobically on usual media, such as Mueller-Hinton agar and also on BCSA. Notably, all strains recovered from patients 2 to 6 were non-pigmented, except for a brown pigmented strain recovered from patient 1 in 2008 and identified as P. pulmonicola by MALDI-TOF MS. The API20NE results were not reliable with an identity score of 58.9 % (T = 0.64) for the species A. xylosoxidans and an identity score of 22.2 % for Alcaligenes faecalis 2 (T = 0.6, biochemical profile: 0040476). Biochemical identification of Pandoraea spp. was not suitable because this genus is not included in the API20NE database. This explains the misidentification of the strain recovered from patient 1 prior to molecular identification. Previous studies have reported the misidentification of P. sputorum as A. xylosoxidans [11] and the misidentification of other Pandoraea species as A. faecalis, A. denitrificans, CDC IV C2 and Acinetobacter spp. [21]. ARDRA, as well as 16S rDNA-based PCRs, resulted in a genus level identification of the Pandoraea sp., but could not discriminate between P. apista and P. pulmonicola. Interestingly, the gyrB gene restriction fragment length polymorphism (RFLP), which had been shown to be a reliable identification method for Pandoraea spp., could not be performed due to amplification failure of the gyrB gene for all the strains. Coenye et al. also reported low amplification of the gyrB gene from the three P. pulmonicola strains tested in their study [15]. Neither the 16S rDNA-based specific PCR proposed by Coenye et al. in 2001 [16] nor the ARDRA test [14] were able to differentiate P. pulmonicola from P. apista. Nevertheless, our 655 bp 16S rDNA gene sequence displayed more identity with the P. pulmonicola type strain LMG 18106 (99.8 %) than with the P. apista type strain LMG 16407 T (98.9 %). Comparing this sequence with other available 16S rDNA Pandoraea strain sequences using the blast2N program (www.ncbi.nlm.nih.gov/blast/bl2seq/wblast2.cgi) resulted in 99.8 % homology with P. pulmonicola LMG 18106 T, 99.7 % with P. pnomenusa LMG 18087 T, 99 % with P. apista LMG 16407 T, 98.9 % with P. sputorum LMG 18819 T, and 98.4 % with P. norimbergensis LMG 18379 T. MALDI-TOF MS found the closest match with reference strain Pandoraea sp [2] 65 RLT (score value > 2 for all six strains). Interestingly, Pandoraea sp [2] 65 RLT displayed 99.8 % sequence identity with Pandoraea pulmonicola (accession number AF139175) and 99.5 % sequence identity with Pandoraea pnomenusa (accession number AY268170) (data provided by Bruker Daltonics). These results were consistent with the 16S rDNA sequence identity obtained above. The results from ARDRA, 16S rDNA sequencing and MALDI-TOF MS indicated that these methods should be used together in order to achieve an accurate identification at the species level.

In our assay, the epidemic strain was resistant to colistin, aminoglycosides, fosfomycin, pefloxacin, ciprofloxacin, ticarcillin, ticarcillin-clavulanic acid, meropenem, ceftazidime, and aztreonam, which was similar to previous reports for other Pandoraea species [8, 10, 11]. Until 2003, all P. pulmonicola strains recovered from patient 1 displayed a more susceptible phenotype with lower MIC values to piperacillin (32 μg/mL), piperacillin-tazobactam (24 μg/mL) and imipenem (3 μg/mL). This susceptibility pattern is similar to the P. sputorum strain described by Fernandez-Olmos et al. [8] with susceptibility to imipenem (MIC value of 4 μg/mL) and piperacillin-tazobactam (MIC value ≤ 16 μg/mL). Notably, patient 1 received piperacillin-tazobactam for the first time in 2007 and imipenem for the first time in January 2009, after the strain became resistant to these antibiotics. After 2003, all recovered strains from patient 1 displayed the same AST pattern as the epidemic strain (resistance to piperacillin, piperacillin-tazobactam and imipenem). Jørgensen et al. already reported intermediate susceptibility to ceftriaxone in P. apista strains [10]. Susceptibility to trimethoprim-sulfamethoxazole was also reported in P. sputorum strains [8].

Analyses of hospitalization periods for patients 1, 2, 3, 4, 5, and 6 are consistent with the chronological acquisition of P. pulmonicola (data not shown). In 2009, the source patient was hospitalized eight times. The clonality of the strains isolated from all six patients was assessed by PFGE analysis. The main hypothesis to explain the bacterial spread between patients is droplet cross-transmission. In our center, all patients have to follow standard precautions and are asked to use an alcohol hand rub. Specific droplet cross-transmission preventive measures consist in the use of respiratory masks for patients chronically colonized with P. aeruginosa and cohorting patients chronically colonized with the B. cepacia complex. In 2009, the number of patients in our center who were chronically colonized with P. aeruginosa was 150 out of 243 (61.7 %). The preventative measures were not strictly followed, and the patients had several contacts without respiratory masks, which is likely to have enabled droplet cross-transmission of P. pulmonicola. However, we cannot completely exclude an indirect droplet mediated cross-transmission or the possibility of exposure to a common bacterial source. Finally, by applying strict cross-transmission prevention measures, i.e. segregation of colonized patients in single hospitalization rooms with use of gowns for medical staff, consultations in specific rooms and on specific days, forbidden access to common areas, use of respiratory masks and use of specific spirometers with single use turbines and mouthpieces, the epidemic spread was controlled. In our study, all patients were chronically colonized with the mucoid P. aeruginosa, and associated pathogens were found in two patients. It would be interesting to determine whether chronic P. aeruginosa colonization is a condition that facilitates the acquisition and colonization of P. pulmonicola. All patients remained chronically colonized with P. pulmonicola after the date of first acquisition, which is consistent with previously described data regarding the ability of this bacterial species to chronically colonize lungs of CF patients. For example, in 2009, P. pulmonicola was recovered from 4/4 sputa (patient 2), 14/14 sputa (patient 3), 10/10 sputa (patient 4), 20/20 sputa (patient 5) and 6/6 sputa (patient 6). As for patient 1, the total number of positive sputa was 50 since the year 2000. The main features for all six patients concerning the clinical outcomes after P. pulmonicola acquisition are summarized in Table 1. Three patients died (patient 1, patient 4 and patient 5). Patient 1 died in October 2010 from severe comorbidities. Patient 4 died in the context of P. aeruginosa bacteraemia (P. pulmonicola was not retrieved in blood cultures). Patient 5 died in May 2012 from multiple-organ failure one month after lung and liver transplantation (blood cultures grew P. pulmonicola, P. aeruginosa, and A. xylosoxidans). Several patients displayed declined lung function (Patient 3, Patient 4 and Patient 5). Two patients are clinically stable (patient 2 and patient 6). Patient 6 had a lung transplantation in January 2012 and is still chronically colonized with P. pulmonicola after bilateral lung transplantation. Altogether, the specific role played by P. pulmonicola in declined lung function is difficult to determine since it is associated with the chronic colonization of P. aeruginosa. However, recent data reported virulence of P. pulmonicola strains, mentioning that P. pulmonicola strains were more virulent than other Pandoraea species and could translocate through polarized lung epithelia with an in vivo virulence comparable to that of B. cenocepacia [21].

Our study notes the role of P. pulmonicola as an emerging pathogen that can cause chronic lung colonization in CF patients. Identification tools need to be accurate and must be based on molecular techniques and MALDI-TOF MS. Cross-transmission preventive measures also need to be strictly followed to avoid an epidemic spread. Hopefully, the use of whole genome sequencing will provide significant additional information concerning the virulence of P. pulmonicola.

Informed consent was obtained from the six patients that were reported in this article. For this retrospective observational study, ethics adherence was not required.