We studied 53 MDR M. tuberculosis strains isolated from different patients between 1993 and 2001 at the Bangui Pasteur Institute. Fourteen of these patients were HIV positive, 30 were HIV negative, and 9 were of undetermined status. Epidemiologic enquiries did not show a social link among patients (unpub. data). A non-MDR, nonbiased control group, which included 263 M. tuberculosis and 2 M. bovis strains, was also studied. These strains included all those from the cohort studied by Espinal et al. (3) for which a subculture was obtained.

Figure A1

Figure A1. Dendrogram of patterns of the 283 strains spoligotyped (53 multidrug-resistant [MDR] and 230 non-MDR strains). For a printer friendly version of Figure A1, available at http://wwwnc.cdc.gov/eid/pdfs/06-0361-FA1.pdf.

All 318 isolates were typed by using the spoligotyping method previously described (13). Spoligotypes were obtained for 283 (53 MDR and 230 non-MDR strains) of the 318 isolates and were analyzed with BioNumerics software (Applied Maths, Kortrijk, Belgium). Dendrograms were constructed according to degree of similarity (Dice coefficient) and comparison with known spoligotypes. Seventy-nine different spoligotypes were identified: 55 included only a single strain, and 24 included 2–56 strains. The 24 clusters were named A to X (Figure A1).

The 53 MDR strains clustered in 8 clusters (Figure, panel A). Twenty-five (47.2%) of 53 clustered in type E, which has characteristics of the T family (ancient M. tuberculosis strains with numerous spacers [14]). The 230 spoligotype patterns of non-MDR strains were grouped into 22 clusters, and spoligotype E was not a major cluster (Figure, panel B). Cluster Q contained the largest number of strains. Its spoligotype is identical to the DB3 pattern ST47 characteristic of the Haarlem family (15). Spoligotypes 97%–99% identical with profiles characteristic of the Haarlem family of strains represent 155 strains. These observations confirm the predominance of the Haarlem type in Africa. However, the Haarlem family was not predominant in our collection of MDR isolates.

The clustering of MDR strains suggested an MDR outbreak; therefore, we looked for other characteristics in cluster E isolates. First, we tested for diversity in the rpoB region, which was likely to be responsible for rifampin resistance. Five variants were found among the 26 MDR strains that constituted clusters E and F (Table): 9 had a Ser-531 (TCG) to Leu (TTG) substitution; 8 and 5 strains contained a substitution of His-526 (CAC) with Tyr (TAC) and Arg (CGC), respectively; 3 had an Asp-516 (GAC) to Val (GTC) variant; and 1 a Leu-533 (CTG) to Pro (CCG) substitution. All these variations are in the rifampin resistance-determining region frequently encountered in strains with a rifampin-resistant phenotype (8). These variants probably determine rifampin resistance and may have occurred independently, not necessarily corresponding to MDRTB transmission, even for strains of the same cluster with the same change in rpoB. Indeed, sociodemographic and epidemiologic characterization of the patients did not show any links between these MDRTB cases. Therefore, rifampin resistance seems to have been acquired independently and repeatedly by cluster E and F strains. To find a way to reduce the dissemination of such strains likely to generate MDR isolates, we characterized strains of cluster E and F. In particular, we looked for single nucleotide polymorphisms (SNPs) in the putative genes mutT1, mutT2, mutT3, Rv3908, mutY, mutM, ada/alkA, and ogt. Sequencing was performed as previously described (11). With reference to published M. tuberculosis sequences, we found 1 synonymous SNP in mutT1 corresponding to Val 265 (GTC) to Val (GTA). This SNP is only present in strains of cluster E and strains 27 (cluster F), 28, and 29 and is absent from all strains in other clusters. Therefore, these MDR strains are characterized by a spoligotype pattern (ST 52 ± spacer 11 or 12 to 15 and ST 107) and the presence of the mutT1 SNP 265.

MDR strains of clusters E and F and strains 28 and 29 (a single difference in spacers between E and F, strain 28 or 29) corresponded to 9 new cases and 19 patients who had received previous treatment. MDR strains in other clusters corresponded to 9 new cases, 15 previously treated patients, and 1 case for which no history was available (3).

We used 2 types of markers to study the genetic diversity of MDR M. tuberculosis strains isolated in Bangui: spoligotyping and SNPs in a series of putative DNA repair genes. Many MDR strains were clustered in 1 spoligotype and carried the same SNP in the anti-mutator gene mutT1. Indeed, 25 of the 53 MDR strains were in cluster E. Thirty-two percent of these MDR strains were from new cases of infection, and 40% were from HIV-infected patients. This cluster was not a major cluster among the 265 non-MDR isolates collected during a 5-month period. The same SNP was found in all strains of cluster E and F tested and in 2 strains that differed by 1 spacer. These strains carry variants of rpoB that confer rifampin resistance, which implies that these strains do not correspond to an MDR-TB outbreak. However, this finding is consistent with the possibility that these strains represent an MDR-prone family, members of which are often associated with MDR phenotypes in Bangui. Detection of strains characterized by the T family spoligotype and mutT1 SNP 265 may be useful to identify patients at risk of developing MDR-TB.