Simultaneous Infection with More than One Strain of Burkholderia pseudomallei Is Uncommon in Human Melioidosis

Abstract

A prospective study was performed to determine the rate at which patients with melioidosis are infected with more than one strain of Burkholderia pseudomallei. Genotyping of 2,058 bacterial colonies isolated from 215 samples taken from 133 patients demonstrated that mixed infection is uncommon (2/133 cases [1.5%; 95% confidence interval, 0.2 to 5.3%]).

Melioidosis is a severe infection caused by the gram-negative bacillus Burkholderia pseudomallei, an environmental organism present across Southeast Asia and northern Australia (4). Despite prolonged antibiotic therapy (≥10 days of intravenous antibiotics followed by 12 to 20 weeks of oral antibiotics), recurrent disease is common (≥6% in the first year) (3, 5). This may be due to relapse caused by B. pseudomallei persisting within a sequestered focus or reinfection by a different strain. Distinguishing between the two is important since clinical trials of oral “eradication” therapy for melioidosis commonly use recurrent disease as a marker for treatment failure (relapse). Furthermore, advice to avoid contact with soil and water following a first episode of melioidosis (not trivial, given that most patients are rice farmers) would be important if reinfection were common.

A previous study performed by us investigated the rate of relapse versus that of reinfection in 116 patients with 123 episodes of recurrent melioidosis (7). We reported that 92 episodes (75%) of recurrent disease were attributable to relapse and 31 episodes (25%) were attributable to reinfection. This was based on a combination of typing techniques in which strains from primary and recurrent infections were initially compared using pulsed-field gel electrophoresis (PFGE), followed by multilocus sequence typing (MLST) when the patterns were discrepant by one or more bands for pairs of primary and recurrent strains from a given patient. The isolates used were taken from a retrospective freezer collection, in which each vial represented six or more colonies from the initial culture plate. A possible criticism of this study is that “reinfection” could actually represent relapse in the event that the primary infection was caused by simultaneous infection with multiple B. pseudomallei strains and different strains were picked by chance during the primary and recurrent episodes. In a previous study of 18 patients presenting to a hospital in northeast Thailand between 1992 and 1993 in which 10 to 40 colonies per patient were picked from primary culture plates, 5 patients (28%) were shown to have infection with more than one genotypic strain when a combination of ribotyping and PFGE was used (8). Of these, four patients had isolates with two genotypic patterns and one patient had isolates with three. Furthermore, patients with polyclonal infection had more-severe disease and poorer outcomes.

The aim of this study was to perform a large, prospective study to define the rate of polyclonal infection in unselected patients presented to a hospital in northeast Thailand, in which multiple colonies were genotyped from multiple samples from a given patient.

We powered the study to compare the mortality rates associated with polyclonal versus single-clone infection. Assuming a polyclonal infection rate of 28% (based on the study reported by Pitt et al. [8], which was undertaken in the same population) and an overall mortality rate of ∼45% in our melioidosis population, at least 102 patients are required to detect mortality rate differences of 70% for polyclonal infection and 35% for clonal infection with 80% power. We determined the number of colonies required from each sample; at least 10 colonies are needed to detect the presence of two strains at a ratio of 1:9 with 85% power. All analyses were calculated using Stata 9.0 (College Station, TX).

A prospective study was conducted between 28 June and 29 September 2006 at Sappasithiprasong hospital, Ubon Ratchathani, northeast Thailand. The study was approved by the Human Research Ethics Committee of the Faculty of Tropical Medicine, Mahidol University, Bangkok. Patients with suspected melioidosis were sought during twice-daily ward rounds of the medical and intensive care wards together with passive surveillance in surgical and pediatric wards. Histories and examinations were performed on patients with one or more clinical features consistent with a diagnosis of melioidosis. Multiple samples were taken from suspected patients within 72 h of admission. A 15-ml blood sample was taken and divided between a BacT/ALERT FA bottle (BioMérieux, NC) for standard culture (5 ml) and an Isolator10 lysis centrifugation tube (Oxoid, Basingstoke, Hampshire, United Kingdom) (10 ml) to allow identification of primary colonies. Samples were taken for culture of sputum/tracheal aspirates, throat swabs, urine, pus, or surface swabs from wounds and skin lesions as appropriate or available. The specimens were hand carried to the onsite research laboratory at the end of each ward round.

The objective of our culture technique was to achieve single B. pseudomallei colonies on primary plates from material taken directly from the clinical sample. The rationale for this was to obtain a true picture of the bacterial population in the primary specimen and to avoid the use of subculture in enrichment broth during which one strain of B. pseudomallei could outgrow a second. The 10-ml blood sample collected into the Isolator tube was processed and spread plated onto Ashdown's agar as previously described (9). Urine samples were spread plated as previously described (6). Samples of pus and respiratory secretions were serially diluted prior to being spread plated on Ashdown's agar plates to obtain single colonies. Throat and wound swabs were streaked directly onto an Ashdown's agar plate. In addition, specimens were cultured using enrichment broth, as previously described (11). All agar plates were examined daily for clinical diagnostic purposes but were maintained at 37°C in air for 4 days before colonies were selected for genotyping. This was because colony morphology variability, a feature that may reflect different strains, is best observed at this time point.

Colonies of presumptive B. pseudomallei were initially identified on the basis of colony morphotype. This included the characteristic colony morphology (purple, flat, dry, and wrinkled) together with six additional colony morphotypes, as described previously (2). Colonies suspected to be B. pseudomallei were tested using an oxidase test and then confirmed using a highly specific latex agglutination test (positive for B. pseudomallei but negative for Burkholderia thailandensis) (1, 12). Colonies on the primary plate were picked for genotyping when available for a given sample; in the event that enrichment broth was positive but primary plates were negative, colonies were picked following subculture. For samples with a single B. pseudomallei colony morphology type present, 10 colonies were picked and saved to independent freezer vials. All colonies were picked when fewer than 10 were present. When more than one colony morphology was present, five of each were picked to independent vials.

Genotyping was performed using a combination of PFGE and MLST. All colonies were subjected to PFGE using SpeI, as previously described (7). Samples from an individual patient were run on the same gel and analyzed using BioNumerics (version 2.5) software (Applied Maths, Belgium). Colonies from a given patient with identical PFGE banding patterns were regarded as genotypically indistinguishable, but isolates with one or more bands different were further examined using MLST, as previously described (10). The alleles at each of the loci were assigned and sequence types defined using the B. pseudomallei MLST website (http://bpseudomallei.mlst.net).

A total of 215 samples were processed from 133 patients with culture-confirmed melioidosis, of whom 8 (6.0%) were less than 15 years of age. Sample types were blood (n = 73), respiratory secretion (n = 54), urine (n = 20), pus (n = 18), throat swab (n = 26), and wound swab (n = 24). Mixed B. pseudomallei morphotypes were observed on 18 of 184 (9.8%) primary plates. Morphotypes I to VII were present in 178 (96.7%), 13 (7.1%), 6 (3.3%), 4 (2.2%), 6 (3.3%), 7 (3.8%), and 3 (1.6%) specimens, respectively.

A total of 2,058 colonies were examined by PFGE. The median number of colonies tested per patient was 10 (interquartile range, 10 to 20; range, 1 to 70). Two patients had only a single colony of B. pseudomallei grown from the Isolator tube, and one patient had 70 colonies obtained from seven specimens (blood, urine, and respiratory secretion and four swabs from pustules at the forehead, hip, and both legs).

A total of 130 patients (97.7%) had colonies with identical PFGE banding patterns, consistent with infection by a single strain of B. pseudomallei. Three patients had samples containing colonies with two different PFGE banding patterns, details of which are summarized in Table 1. MLST was performed on a representative of each banding pattern from each sample. This demonstrated that two of the three patients were each infected with two different strains of B. pseudomallei (confirming polyclonal B. pseudomallei infection in 2/133 cases (1.5%; 95% confidence interval, 0.2 to 5.3%). The third patient had multiple samples containing strains that differed by two bands on PFGE, which on MLST was shown to be a single clone (Table 1). The two patients with polyclonal infection survived to hospital discharge, completed 20 weeks of oral antibiotic treatment, and were well on the 3-month follow-up.

TABLE 1.

Details of three patients with melioidosis infected with more than one strain of B. pseudomallei

Patient no.	Sample type	Colony typea	No. of colonies tested	PFGE pattern(s) (no. of colonies)	No. of bands of difference between indicated PFGE patterns	MLST for indicated PFGE patterns	Clinical features and outcome
1	Respiratory secretions	I	10	A (9), B (1)	12 (A vs B)	A, ST207 B, ST9	Female, 3 yrs old, presented with right parotid abscess and febrile convulsion; survived
2	Throat swab	I	10	C (10)	11 (C vs D)	C, ST167	Female, 55 yrs old, diabetes presented with acute pneumonia and hypotension; survived
	Respiratory secretions	I	10	C (9), Db (1)		D, ST511
3	Blood	I	9	Db (9)		D, ST511	Female, 30 yrs old, presented with acute pneumonia and acute hyperglycemia; survived
	Throat swab	I	10	D (9), E (1)	2 (D vs E)	E, ST511
	Respiratory secretionsc	I	5	D (4), E (1)
		II	5	D (4), E (1)
		IV	5	D (5)

^aAs described in reference 2.

^bPatients 2 and 3 were both infected with B. pseudomallei ST 511; they were not admitted during the same period, and contamination of the sample taken from patient 2 with a strain from patient 3 is considered highly unlikely.

^cThree colony types were detected in one respiratory secretion sample from patient 3.

This large prospective study of unselected patients with melioidosis has demonstrated that primary infection with more that one strain of B. pseudomallei is very uncommon. We cannot exclude the possibility that in vivo selection of a biologically successful clone occurs at the primary site of infection following inoculation of more than one B. pseudomallei strain, but our strategy of culturing all available sample types from a given patient reduces the probability of missing mixed infection in the primary site and clonal spread to secondary sites. Our findings differ considerably from those of Pitt et al. (8). The reason for this is not clear; laboratory contamination leading to more than one strain in a sample or freezer vial is unlikely since the PFGE patterns were not shared between isolates associated with mixed infection, although this does not rule out the possibility. Case and sample type selection could lead to variable results between the two studies. Pitt et al. reported mixed infection involving samples collected during the months of August and September of 1992 and 1993, compared with our strategy of recruiting consecutive patients between June and September. The distributions of sample types were also different, with 14/18 (78%) samples in the study by Pitt et al. representing blood cultures compared with 73/215 (34%) in this study.

We suggest that our previous study in which we reported that 25% of cases of recurrent melioidosis were due to reinfection represented a true measure and that patients who survive to hospital discharge following a first episode require counseling regarding the risk and strategies for avoidance of further, independent episodes of melioidosis.