Analysis of Mycobacterium avium Complex Isolates from Blood Samples of AIDS Patients by Pulsed-Field Gel Electrophoresis

Mycobacterium avium complex (MAC) is the commonest cause of systemic bacterial infection in patients with AIDS, with an annual incidence of 20% before the widespread use of highly active antiretroviral therapy (8). The development of disseminated MAC infection produces significant morbidity and is an independent predictor of death (6). MACs are found in water or earth and are usually opportunistic pathogens causing disease in an immunocompromised host. There is no evidence that MAC infection occurs as a result of person-to-person transmission. If such transmission does occur, it would have implications for prophylaxis and other strategies to prevent MAC infection in AIDS patients.

A number of studies have shown that MAC strains can be distinguished by restriction digest and pulsed-field gel electrophoresis (PFGE) (4, 7, 11, 12). Analysis of large DNA restriction fragment (LRF) patterns generated by PFGE of MAC has demonstrated that unrelated strains of MAC are polymorphic whereas strains from the same patient (isolated either from different sites or at different time points) tend to have identical patterns (7). This makes LRF analysis attractive for strain typing of MAC for epidemiological studies. In the present study, the LRF patterns of initial MAC blood isolates (collected for the Canadian HIV Trials Network MAC Treatment trial [10]) of patients were compared to determine the prevalence of clustering of strains in this population and whether there was any evidence of person-to-person transmission.

The Canadian HIV Trials Network MAC Treatment trial was a multicenter collaborative randomized treatment study of AIDS patients with MAC bacteremia (10). Patients were randomized into a treatment protocol after enrollment and monitored with repeat blood cultures at 2, 4, 8, 12, and 16 weeks. A total of 187 patients were eligible for evaluation for the study. All isolates were sent to the Provincial Laboratory of Northern Alberta for species confirmation and susceptibility testing and were stored at −80°C. For this study, coded isolates were grown onto solid media from frozen stock and subcultured from a single colony and a suspension was embedded in an agarose plug. Bacterial cells in the plug were lysed with lysozyme (1 mg/ml final concentration) and RNase (20 mg/ml final concentration) and incubated at 37°C for 4 days. Genomic DNA was digested with AsnI and then subjected to PFGE by the method of Slutsky et al. (11), resulting in LRF patterns. Each PFGE gel incorporated two lanes of lambda ladder molecular mass markers (catalog no. N0340S; New England Biolabs, Missasauga, Ontario, Canada) (range, 48.5 to >1,000 kb). Ethidium bromide-stained images were photographed, printed, and digitalized with the IMAGER video camera system (Appligene, Illkirch, France). Digitalized gel images were analyzed using GelcomparII computer software (Applied Maths, Kortrijk, Belgium). All isolates determined by computer analysis to be indistinguishable or similar were examined by visual comparison of the corresponding lanes on the matched original pictures.

Of 187 MAC initial blood isolates, 2 could not be regrown after frozen storage. Therefore, the remaining 185 initial MAC blood isolates were processed for LRF analysis. Technically acceptable patterns were produced for 177 of the isolates. These 177 coded MAC isolates from 10 Canadian cities were evaluated by blind comparisons of LRF patterns. A representative sample of LRF patterns is shown in Fig. 1. Two or more isolates having indistinguishable patterns were designated a cluster. After identification of clusters, the samples were unblinded and epidemiological data on the patients within clusters were obtained. It was found that patients with isolates belonging to the same cluster were geographically related but not otherwise epidemiologically linked. Of seven clusters that were identified, six consisted of isolates from patients living in the same city as the others with isolates in the same cluster and one consisted of isolates from patients living in cities near those of others with isolates in the same cluster (Table 1). Cluster II consisted of isolates from two patients, one from Montreal and the other from Sherbrooke, a city 150 km away. A numerical simulation by the Monte Carlo method (13) was performed to determine the probability due to chance alone of isolates in clusters being geographically related. The frequency of occurrence due to chance of the isolates in six of the seven clusters originating from patients in the same city would be 8 per 100,000. This would suggest that there is either common source exposure or person-to-person transmission. In this study, there was no suggestion for person-to-person transmission in patients within clusters in that the cases were not epidemiologically closely related. This finding is in keeping with a study by Bauer et al., who reported that when partners had MAC infection, the strains isolated from each partner were different (3).

Arbiet et al. have demonstrated that 15% of patients may be bacteremic with more than one strain of M. avium simultaneously (1), whereas Picardeau et al. found that less than 5% of patients had polyclonal bacteremia (9). In this study, we chose a single colony and hence a single strain to type by PFGE for each patient. If more than one strain were present, we would miss any additional strains. Therefore, we may have underestimated the number of clusters if patients had another strain which matched the strain of another patient. However, as the colony for each isolate was chosen at random, it is unlikely that the overall conclusions of the study would be affected.

The source of MAC infections is thought to be environmental, with water being an important reservoir. Von Reyn et al. isolated MAC prospectively and found that two clusters of MAC strains from patients were linked to strains isolated from hot water of the hospital that the patients attended (12). Similarly, Aronson et al. described a close relationship between MAC isolates from potable water and clinical isolates (2). Bauer et al. found that some strains of MAC were isolated from patients that were geographically widespread in Denmark and could be found in peat used for potting soil (3). This suggested that peat was the environmental source of MAC, particularly when MAC could not be found in local potable water. Chin et al. have demonstrated that colonization in either lung or stool by MAC correlates with later invasive disease (5). Presumably, MAC is acquired from the environment, colonizing patients and then causing invasive disease, usually when immunity is compromised.

The series of MAC blood isolates employed in the present study is to date the largest to have been analyzed by a DNA typing method, and the results of the analysis lend further support to the premise that MAC is acquired from the environment and that clustering is likely related to common exposure to an environmental source.

• Open in new tab
• Download powerpoint

FIG. 1.

Sample of LRF patterns of M. avium isolates (lanes 1 to 11). Lanes MW, lambda ladder molecular mass markers (in kilobases).

• Copyright © 2003 American Society for Microbiology