During January 2007–December 2008, health institutions in Valle del Cauca sent M. tuberculosis isolates to the Centro Internacional de Entrenamiento e Investigaciones Médicas in Cali, Colombia, for drug susceptibility testing. A total of 324 M. tuberculosis isolates from individual patients (new and previously treated) were subjected to first-line drug susceptibility testing by using the agar proportion method with 7H10 medium.

In 2009, with approval from the Centro Internacional de Entrenamiento e Investigaciones Médicas Institutional Review Board, we thawed 160 (49%) of 324 isolates (76 MDR and 84 drug susceptible) and cultured them on Löwenstein-Jensen agar slants. Isolates with other susceptibility profiles were excluded from this analysis. Isolates were obtained from 9 municipalities in Valle del Cauca: Buenaventura (n = 113), Cali (n = 36), and 7 other locations (n = 11). Eighty-seven patients (54.4%) had new cases. There were more male (57.5%) than female patients, and the median age of all patients was 32 years (range 1–82 years).

DNA extraction was performed by using the hexadecyltrimethylammonium bromide method (11), and isolates were subjected to spoligotyping (12). Spoligopatterns obtained were analyzed independently by 2 readers and compared with SpolDB4.0 (Pasteur Institute of Guadeloupe, Les Abymes, Guadeloupe) and MIRU-VNTRplus (www.miru-vntrplus.org) to assign them to a known genotype family.

Statistical analysis was conducted by using Stata version 9.0 (StatCorp LP, College Station, TX, USA) and the Openepi (www.openepi.com). Odds ratios (ORs) and 95% confidence intervals were calculated by using the Fisher exact test method. Chi-square test was used to determine statistical significance.

Thirteen genotypes were identified among the 160 isolates: LAM9 (32; 20%), H1 (32; 20%), Beijing (25; 15.6%), T1 (9; 5.6%), U (9; 5.6%), LAM2 (6; 3.8%), LAM3 (5; 3.1%), LAM1 (4; 2.5%), X1 (4; 2.5%), H3 (4; 2.5%), U (LAM30) (2; 1.3%), X3 (2; 1.3%), and LAM6 (1; 0.6%). Orphan genotypes accounted for 25 (15.6%) isolates, and 12 of these isolates were grouped in 4 patterns (Table 1).

Overall, the LAM and H familes were the most common among isolates evaluated, particularly among susceptible isolates. H strains represented a homogeneous group and were distributed in 4 spoligo-international types (SITs 47, 49, 50, and 62). LAM isolates were distributed in 10 SITs (17, 20, 42, 64, 130, 162, 469, 545, 1711, and 1803).

Three results were obtained regarding the Beijing genotype. First, a Beijing family strain (SIT 190) caused the largest cluster among the MDR isolates, comprising 24 cases. The remaining Beijing strain (SIT 1) corresponded to a susceptible isolate from Buenaventura (Table 1). Second, the Beijing genotype showed a strong correlation with female patients and patients residing in Buenaventura (Table 2). Third, Beijing SIT 190 was found in 2 of 4 XDR TB isolates, as confirmed in susceptibility testing by the Instituto de Salud Pública de Chile (Santiago, Chile). These 2 isolates were found in 2 women, 16 and 24 years of age. The other 2 XDR TB isolates were an H strain and an orphan genotype strain.

Although our study used a convenience sample, it identified a high frequency of Beijing strains among MDR TB isolates and showed an association between the Beijing genotype and MDR TB in Latin America. In Buenaventura, where a high rate of primary drug resistance has been observed (3,6), the Beijing genotype is associated with MDR TB and thus transmission. Moreover, given the limited number of MDR isolates tested, the emergence of Beijing strains in Colombia may already be much larger than what we observed.

In multiple areas worldwide, the Beijing genotype has been associated with young patients and active and recent transmission (8). In our study, the same tendency was observed, and this may suggest the emergence of this genotype family in Colombia.

The high frequency of the Beijing strain might be caused by bacteriologic factors, host factors, or both (8). Valle del Cauca has a high proportion of persons of African descent (27.2% according to the national census in 2005); especially in Buenaventura, where 23 of the 25 Beijing isolates were found. Also, Buenaventura, which has >300,000 inhabitants, has a high TB incidence (72 cases/100,000 inhabitants/year) (13). Commercial and tourism activities and high population mobility in this city may contribute to dissemination of the Beijing genotype to other regions of Colombia. Two persons infected with the Beijing strain found in Cali, Colombia, in this study and a recently described 15-year-old person with MDR TB infected with a Beijing-like genotype, who died in Bogotá shortly after initiation of treatment, may represent preliminary evidence of the mentioned risk (14).

Our results differ from those of a report that described a low frequency of Beijing strains in Latin America and no association with drug resistance (10). However, Colombia might have recently become a port of entry for Beijing strains and this entry may have started earlier without being detected. Additional epidemiologic and clinical information is necessary to correlate our findings with other factors, such as M. bovis BCG vaccination, ethnic group, disease severity, and outcome.

More discriminative typing methods such as mycobacterial interspersed repetitive unit–variable number tandem repeat analysis (15) or whole genome sequencing would enable typing to the strain level. This typing would shed light on epidemiologic links between the cases we reported and worldwide spread of the Beijing strain and on the location of these Beijing strains in a worldwide phylogenetic tree. Nevertheless, spoligotyping used in this study is sufficient to conclude that drug-resistant Beijing strains have become a public health problem in Buenaventura, Colombia.