Direct microscopic examination of specimens showed substantial presence of PNNs and gram-negative short bacilli, except in the bronchoalveolar lavage fluid from case-patient 2, in which no bacteria was observed. In both case-patients, colonies were apparent after 24 or 48 hours of incubation (37°C, 5% CO₂) on PolyViteX chocolate agar. In case-patient 2, they also grew on trypcase soy agar with 5% horse blood and Haemophilus chocolate agar with bacitracin (bioMérieux). The colonies were medium sized (1–2 mm), smooth, round, convex, and grayish; those from case-patient 2 were very mucoid. Microscopic examination of colonies showed gram-negative short bacilli. Identification was inconclusive with Vitek 2 system IdGN cards (bioMérieux) and RapID NH (Remel, Lenexa, KS, USA). The Api 20NE strip (bioMérieux) identified B. avium based on the score 0000067 with high percentage (96.7%) and typicality (T = 1). Routine identification by matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry (Brucker Daltonik GmbH, Germany) gave B. hinzii with good scores (T = 2.29 for case-patient 1 and T = 2.35 for case-patient 2). Final identification for both cases involved amplification and sequencing of 1,480 nt of 16S rRNA gene, then use of the blastn program (http://www.ncbi.nlm.nih.gov/BLAST/) and the GenBank database. The nucleotide sequences showed the best BLAST hit with B. hinzii (99.9% homology in both cases). Antimicrobial susceptibility testing was performed on Mueller-Hinton medium by an Etest method (bioMérieux) and MICs determined (Table).

B. hinzii is found in the respiratory tract of poultry. Although it was initially described as a commensal bacterium in birds, some veterinary isolates appear to have pathogenic properties: Register et al. reported that it is associated with tracheal lesions in turkeys (1,6). B. hinzii has been sporadically described in rabbits (6) and rodents (2) and had been reported to cause pulmonary disease in laboratory mice (7) and bacteremia in rats (2).

Few cases have been described in humans. Most involved bacteremia or respiratory or digestive infections (3–5). B. hinzii was first isolated in 1957 from the sputum of a patient in France but was misidentified as Alcaligenes faecalis (8). In 1994, Cookson et al. described B. hinzii bacteremia in an AIDS patient (4). Two other cases of bacteremia in immunodeficient patients were associated with the isolation of B. hinzii from the pulmonary tract (9,10).

Two digestive infections have been reported. A fatal case was described in an immunocompetent man (11) with cholestasis and bacteremia. The second case was a biliary infection in a liver transplant recipient: B. hinzii was isolated from 4 bile samples during a 6-month period (3), demonstrating that colonization by B. hinzii can be long-lasting.

B. hinzii appears as an opportunistic pathogen causing respiratory infections in cystic fibrosis patients (5,12). Other respiratory infections in both immunodeficient and immunocompetent persons have been reported. Gadea et al. isolated B. hinzii associated with Nocardia asteroides in a bronchoalveolar lavage sample from an AIDS patient (13). Palacian Ruiz et al. described a B. hinzii respiratory infection in an immunocompetent elderly woman that was associated with Klebsiella oxytoca (14). B. hinzii was isolated 8 times during a 1-year period from the respiratory tract of a cystic fibrosis patient (5).

B. hinzii infections are presumably underdiagnosed because of misidentification by the routine phenotypic identification procedures that are the basis for Phoenix (Phoenix BD Diagnostic Systems, Sparks, MD, USA) and Vitek (bioMérieux) automated systems and for the API 20NE strip manual system (bioMérieux). Indeed, B. hinzii is not referenced in the API 20NE database and often gives the same score as B. avium. The API 20NE database should be updated and the score 0000067 should be referred to as B. avium–B. hinzii complex. The mean score for 6 clinical isolates analyzed with the MALDI-TOF mass spectrometry system (Brucker Daltonik GmbH) by the French National Reference Centre of pertussis and other bordetelloses was T = 2.31, with T = 2.063 as the minimal score. MALDI-TOF is useful to discriminate B. hinzii from other Bordetella species (15). Sequencing the 16S rRNA gene is the most reliable technique to confirm the species, but the MALDI-TOF system is more suitable for routine identification and would enable more cases to be detected.

Although poultry seems to be the major reservoir (1), we cannot exclude the possibility that mammals such as rabbits and rodents also could be potential reservoirs (2,7). Humans can become infected by aerosols from the avian reservoir, which probably was the route for case-patient 1, who had a pulmonary infection long after exposure. Survival in the digestive tract is another specificity of B. hinzii among the Bordetella species. This bacterial survival was illustrated for case-patient 2, who had a positive culture from a rectal swab sample; the bacteria most likely was transmitted by the oral route, possibly after ingestion of contaminated poultry products (3) or deglutition of respiratory secretions.

Several reports have demonstrated the prolonged persistence of B. hinzii in the respiratory and digestive tracts (3,5); such persistence, at least in the respiratory tract, is also observed for B. petrii and B. bronchiseptica (16). This persistence may explain why infection can develop long after exposure and makes identification of the source difficult.

The characteristics of all the reported cases suggest that B. hinzii is an opportunistic pathogen in humans. Other pathogens were isolated in several cases such that the extent of its pathogenicity remains obscure. Surprisingly, Staphylococcus sp. was the most frequently associated pathogen, raising the possibility of synergy between these 2 bacteria (5). The role of B. hinzii is still unclear in immunocompetent patients, where it may act as colonizer. For case-patient 2, the mucoid aspect of the colonies is in accordance with prolonged carriage, as described for Pseudomonas aeruginosa (17).

Treatment of nonclassical Bordetella infections is not standardized. The interpretation of antimicrobial sensitivity testing is not established and is usually done by inference from other nonfermentative gram-negative rods. According to in vitro sensitivity testing for human isolates described in the literature and pharmacology, piperacillin/tazobactam and carbapenems (excluding ertapenem) may be effective (Table). Optimal duration of treatment has not been established but should be long enough to cure the infection and, if possible, eliminate the bacteria. Both patients reported here were initially treated with piperacillin/tazobactam without clinical improvement despite the sensitivity of the isolates to this association; presumably, the treatment duration was too short. The curative doses needed to eliminate B. hinzii appear to be high because B. hinzii was isolated from case-patient 1 despite prophylactic treatment with trimethoprim/sulfamethoxazole, a combination to which the isolate was susceptible. Numerous discrepancies exist between the results of disk diffusion and MIC tests (5), as observed for B. bronchiseptica (A. Ce Coustumier, unpub. data). MIC testing should be performed to confirm the sensitivity of any such isolates. We can speculate about possible antimicrobial resistance acquisition, regarding fluoroquinolone resistance in case-patient 1 and his exposure to this pharmacologic class during previous hospitalizations, as described elsewhere (3), but data remain insufficient to prove it. Microbiological tests after the episode are required to evaluate the effectiveness of treatment.

In conclusion, we report here 2 cases of B. hinzii pulmonary infection in immunodeficient patients, probably after avian exposure. Although the transmission could not be clearly established, a potential link exists between the occupational exposure and the isolation of B. hinzii in the pulmonary tract. The respiratory or digestive carriage can be prolonged such that the infection might emerge only long after contamination, making identification of the source difficult. Although B. hinzii is well established as opportunistic, microbiologists and clinicians need to be aware of the difficulty in diagnosing infections by this species using routine methods. Identification of all clinical isolates belonging to nonclassical Bordetella sp. should be confirmed by a reference laboratory. Further clinical and microbial investigations are necessary to understand the epidemiology and the pathogenicity of B. hinzii. Optimal antimicrobial treatments need to be established and supported by pharmacology and antimicrobial in vitro sensitivity testing (bacteriostasis, bactericidy).