ABSTRACT
Recent genotyping studies of Mycobacterium tuberculosis in Ethiopia have reported the identification of a new phylogenetically distinct M. tuberculosis lineage, lineage 7. We therefore investigated the genetic diversity and association of specific M. tuberculosis lineages with sociodemographic and clinical parameters among pulmonary TB patients in the Amhara Region, Ethiopia. DNA was isolated from M. tuberculosis-positive sputum specimens (n = 240) and analyzed by PCR and 24-locus mycobacterial interspersed repetitive unit–variable-number tandem-repeat (MIRU-VNTR) analysis and spoligotyping. Bioinformatic analysis assigned the M. tuberculosis genotypes to global lineages, and associations between patient characteristics and genotype were evaluated using logistic regression analysis. The study revealed a high diversity of modern and premodern M. tuberculosis lineages, among which approximately 25% were not previously reported. Among the M. tuberculosis strains (n = 138) assigned to seven subgroups, the largest cluster belonged to the lineage Central Asian (CAS) (n = 60; 26.0%), the second largest to lineage 7 (n = 36; 15.6%), and the third largest to the lineage Haarlem (n = 35; 15.2%). Four sublineages were new in the MIRU-VNTRplus database, designated NW-ETH3, NW-ETH1, NW-ETH2, and NW-ETH4, which included 24 (10.4%), 18 (7.8%), 8 (3.5%), and 5 (2.2%) isolates, respectively. Notably, patient delay in seeking treatment was significantly longer among patients infected with lineage 7 strains (Mann-Whitney test, P < 0.008) than in patients infected with CAS strains (adjusted odds ratio [AOR], 4.7; 95% confidence interval [CI], 1.6 to 13.5). Lineage 7 strains also grew more slowly than other M. tuberculosis strains. Cases of Haarlem (OR, 2.8; 95% CI, 1.2 to 6.6) and NW-ETH3 (OR, 2.8; 95% CI, 1.0 to 7.3) infection appeared in defined clusters. Intensified active case finding and contact tracing activities in the study region are needed to expedite diagnosis and treatment of TB.
INTRODUCTION
Tuberculosis (TB) has been a significant cause of illness and death among humans for centuries. TB is currently responsible for more than 8.6 million new cases and 1.3 million deaths annually (1). The World Health Organization (WHO) stated in 2012 that the incidence, prevalence, and mortality from TB declined significantly in the preceding 10 years. However, progress in contracting TB is being fuelled by the emergence of multidrug-resistant (MDR) and extensively drug-resistant (XDR) TB and due to a large reservoir of infectivity in TB-positive individuals who remain undetected. An improved understanding of the epidemiology, pathogenesis, and genetic diversity of the causative agent Mycobacterium tuberculosis is crucial for better control of the disease.
M. tuberculosis strains are presumed to have coevolved with ancient hominids and continue to evolve today. Based on bioinformatic analysis of mycobacterial genomes, M. tuberculosis strains cluster into seven lineages, each associated with specific global geographical locations (2). These lineages are as follows: lineage 1, Indo-Oceanic; lineage 2, East Asian, including “Beijing”; lineage 3, Central Asian (CAS)/Delhi; lineage 4, Euro-American, including the Latin American-Mediterranean (LAM), Haarlem, X type, and T families; lineage 5; lineage 6, West African 1 and West African 2, respectively; and lineage 7 (2). Lineage 2 (Beijing) is highly prevalent in East Asia, lineage 3 is widely distributed in the Indian subcontinent and some countries in East Africa, lineage 4 is commonly found in Europe, the Middle East, America, and some parts of Africa, lineages 5 and 6 are localized in West Africa, and lineage 7 has recently been reported in Ethiopia and among Ethiopian immigrants in Djibouti (2).
Clinical and epidemiological data reveal that Beijing and Indo-Oceanic strains of lineage 2 are associated with more co-dissemination of TB and meningitis than strains from the Euro-American lineage (3). The Beijing genotype also frequently acquires drug resistance and is associated with high bacillary load in acid-fast bacillus (AFB) smears (4). The East African Indian lineage tends to be transmitted at a lower rate than other lineages (5). Mycobacterium africanum acquires drug resistance at a lower rate than the Euro-American lineage in Ghana (6).
While the genetic diversity of M. tuberculosis lineages in Ethiopia has been investigated (7–11), there is little or no data correlating sociodemographic and clinical factors with M. tuberculosis lineages. This study investigates the relationships between the clinical presentation of pulmonary TB (PTB) patients in Amhara Region, Ethiopia, and infection with specific M. tuberculosis strains. Clinical specimens and case data were collected between 2008 and 2010, and characterization by spoligotyping identified a high frequency of infection with lineage 7 M. tuberculosis international spoligotypes (SIT) 910 and 1729 (12). The analysis of these clinical data is extended to include description of the diverse phylogeny of M. tuberculosis strains infecting the patient group and correlation of clinical parameters specific to M. tuberculosis lineage 7 strains inducing a longer patient delay in seeking treatment.
MATERIALS AND METHODS
Design and population.Clinical and sociodemographic information and sputum specimens (n = 240) were collected from recently diagnosed PTB patients in a cross-sectional study. Data collection sites included selected hospitals and health centers in the study region of Amhara, Ethiopia. Patient delay in seeking treatment is defined as the period from the start of TB symptoms until the first presentation to a medical provider. The study was approved by regional ethical review committees in Norway and Ethiopia. Informed consent was obtained from all subjects.
Statistical analyses.The data were entered and analyzed using the Statistical Package for the Social Sciences (SPSS) version 20 (SPSS, Chicago, IL). Group differences were tested using the Mann-Whitney test. Univariate and multivariate logistic regression analyses were used to identify association between predictors and outcome variables. A two-sided P value of <0.05 was considered statistically significant.
Laboratory methods. (i) Smear microscopy and sputum culture.Sputum smear microscopy for acid-fast bacilli (AFB) was done using the Ziehl-Neelsen staining technique. Sputum culture was performed using conventional Löwenstein-Jensen (LJ) slants (13). HIV testing was performed on samples from all participants following the standard procedure in Ethiopia.
(ii) Cultivation of M. tuberculosis.Clinical M. tuberculosis lineage 7 strains were cultured in the semiautomated Bactec mycobacterial growth indicator tube (MGIT) 960 system (BD Biosciences), in modified Middlebrook 7H9 broth supplemented with oleic acid-albumin-dextrose-catalase (OADC) and the antibiotics polymyxin B, amphotericin B, nalidixic acid, trimethoprim, and azlocillin (PANTA). Growth of M. tuberculosis lineage 7 and other isolates was studied by cultivating the strains on Middlebrook 7H10 agar plates for 5 weeks, followed by measurement of colony diameter and dry weight.
DST assays.Drug susceptibility testing (DST) was performed by the proportional absolute concentration (14) and Bactec mycobacteria growth indicator tube (MGIT) (Becton Dickinson) methods (15) following the manufacturer's instructions.
DNA isolation and spoligotyping.DNA was isolated from M. tuberculosis-positive cultures according to the standard procedure (16). Species identification by PCR targeting the RD9 region was performed as previously described (17). The sputum samples were subjected to spoligotyping according to Kamerbeek et al. (18).
MIRU-VNTR.DNA was prepared from 240 M. tuberculosis cultures. The DNA extracts from nine isolates did not provide adequate PCR products and were excluded. The mycobacterial interspersed repetitive-unit–variable-number tandem-repeat (MIRU-VNTR) analysis was performed at the Norwegian Institute of Public Health, Oslo, Norway. DNA samples prepared from 231 M. tuberculosis strains were analyzed by the standardized 24-locus MIRU-VNTR assay as previously described (19). In brief, 24 MIRU-VNTR markers were amplified by PCR using the MIRU-VNTR typing kit (Genoscreen, Lille, France). DNA fragments were separated by capillary electrophoresis using an automated 48-capillary ABI-3730 DNA analyzer (Applied Biosystems, Foster City, CA). PCR fragments were sized, and MIRU-VNTR alleles were assigned using the GeneScan LIZ1200 size standard and customized GeneMapper software package version 3.7 (Applied Biosystems). The results are presented as a set of digits, the MIRU code, that reflect the number of MIRU sequence repeats.
Phylogenetic analyses.The MIRU-VNTRplus online web application was used to classify the M. tuberculosis lineages identified in this study (http://www.miru-vntrplus.org/MIRU/index.faces) In brief, the MIRU-VNTR 24 loci and spoligotyping patterns were uploaded, and the “similarity” search option was selected to classify lineages based on the best match with the reference strains in the database. A categorical coefficient of 1 and distance cutoff of 0.17 that corresponds to a 4-locus difference were chosen. Thereafter a phylogenetic tree-based classification was applied for strains that were not identified by the best-match approach. A dendrogram was generated using the unweighted-pair group method using average linkages (UPGMA). To confirm the lineage classification, a minimum-spanning tree (MST) was calculated based on the MIRU-VNTR 24-locus data. A cluster was defined as two or more strains with similar genetic patterns. Strains that did not yield identical genetic patterns were referred to as “nonclustered.” The recent transmission index was calculated as the number of clustered patients minus the number of clusters divided by the total number of patients.
Operational definition of lineage 7 isolates.Lineage 7 isolates are defined by several criteria, including (i) a deletion from the 4th to 24th spoligotyping spacers, (ii) a high degree of similarity across the entire genome with other lineage 7 strains (9), and (iii) the presence of 2 or more repeats at the 24th MIRU-VNTR loci.
RESULTS
Study population.DNA samples (n = 231) were prepared from sputum samples from M. tuberculosis-positive patients. The patient group included 126 (54.5%) males and 109 (45.5%) females. The mean age was 30 years. Fifty-one (22.1%) of the study subjects were HIV positive.
Distribution of lineages.Using the “identification by similarity” search option of the MIRU-VNTRplus web application, 118 (51.0%) of the strains were assigned to lineages already represented in the database. The majority, 60 (26.0%) strains, were of the Central Asian (CAS) family, followed by 36 (15.6%) lineage 7 strains and 35 (15.2%) Haarlem strains. Approximately half (48.9%) of the isolates (n = 113; 48.9%) did not match any of the lineages described in the database. Phylogenetic tree analysis of these strains identified four novel clusters, which were designated NW-ETH3, NW-ETH1, NW-ETH2, and NW-ETH4. These new clusters comprised 24 (10.4%), 18 (7.8%), 8 (3.5%), and 5 (2.2%) strains, respectively. Twenty-three (10%) isolates could not be assigned to any of the available or new lineages (Fig. 1 and 2). To verify lineage classification, a minimum-spanning tree (MST) was derived (Fig. 1). All lineages detected by the UPGMA analysis were also identified as clonal complexes in the MST (Fig. 2), and thus the deduced phylogeny of these M. tuberculosis strains was validated.
UPGMA tree of Mycobacterium tuberculosis strains based on MIRU-VNTR 24-locus copy numbers. Abbreviations: CAS, Central Asian strain; LAM, Latin American-Mediterranean.
Minimum-spanning tree (MST) of Mycobacterium tuberculosis strains based on MIRU-VNTR 24-locus copy numbers. The M. tuberculosis clonal complexes are represented by different colors. Circle size is proportional to the number of MIRU-VNTR types belonging to each complex. Abbreviations: CAS, Central Asian strain; LAM, Latin American-Mediterranean.
Patient delay in seeking treatment and M. tuberculosis lineage association.For the complete data set, the median self-reported patient delay was 21 days (interquartile range [IQR], 7 to 60 days). However, the median duration of first seeking health care for patients infected with lineage 7 strains was significantly higher than median duration of seeking health care for all patients (Mann-Whitney test, P < 0.008) (Fig. 3). Patients infected with lineage 7 strains were highly likely to delay seeking care compared to patients infected with CAS family strains which is the predominant lineage in our data (adjusted odds ratio [AOR], 4.7; 95% confidence interval [CI], 1.6 to 13.5) (Table 1).
Patient delay associations with infection by non-lineage 7 and lineage 7 Mycobacterium tuberculosis strains. P < 0.008.
Association between sociodemographic/clinical factors and patient delay in seeking treatmenta
MIRU-VNTR patterns and clustering.A high discriminatory power was obtained by a combination of MIRU-VNTR and spoligotyping. A total of 157 distinct MIRU-VNTR genotype patterns were observed. Seventy-four (32%) isolates shared a genotype pattern with at least one other isolate and were grouped in 22 clusters comprising 2 to 11 isolates. Spoligotyping identified in 65 distinct typing patterns grouped into 26 clusters, comprising 2 to 34 isolates (12). The recent transmission index and clustering rates were 22.5% and 32%, respectively. Cluster 12 was the largest cluster, with 11 strains of NW-ETH3; the multilocus variable-number tandem-repeat analysis (MLVA) M. tuberculosis complex (MTC) 15-9 type for this strain was 594-15. The second largest cluster consisted of the CAS/Delhi family (n = 7; cluster 7, MLVA MTC 15-9 type, 1557-32). The third largest cluster was Haarlem (n = 5; cluster 9, MLVA MTC 15-9 type, 291-31).
M. tuberculosis strain clustering and drug resistance.The odds of clustering for streptomycin-resistant strains was 2.9 times higher than that for streptomycin-sensitive strains (OR, 2.9; 95% CI, 1.2 to 7.1). Strains with any drug resistance to first-line anti-TB drugs (OR, 2.1; 95% CI, 1.1 to 4.3) exhibited a higher degree of clustering than sensitive strains. Haarlem (OR, 2.8; 95% CI, 1.2 to 6.6) and NW-ETH3 (OR, 2.8; 95% CI, 1.0 to 7.3) strains were more likely to cluster than CAS family strains (Table 2).
Association between sociodemographic/clinical factors and M. tuberculosis strain clustering
The overall frequency of resistance to any drug tested among the isolates was 16.9% (n = 39). The proportions of mono-drug-resistant, poly-drug-resistant, MDR, and MDR+ TB isolates (where poly drug resistance is resistance to more than one first-line anti-TB drug, MDR indicates resistance to rifampicin and isoniazid, and MDR+ indicates resistance to rifampicin and isoniazid plus resistance to any other first-line anti-TB drugs) were 12.6%, 3.9%, 0.9%, and 0.9%, respectively (Table 3). Using the CAS family as a reference, NW-ETH3 (OR, 3.4; 95% CI, 1.1 to 10.2) isolates were more likely to be resistant to any first-line anti-TB drug. Haarlem strains (OR, 5.1; 95% CI, 1.3 to 20.9) were more likely to be resistant to streptomycin (Table 4). In contrast, Haarlem isolates were more resistant to any of the first-line anti-TB drugs and streptomycin than any other lineage group (OR, 2.5, and 95% CI, 1.1.1 to 5.7, and OR, 3.5, and 95% CI, 1.4 to 9.3, respectively) (Table 5).
Drug resistance stratified by Mycobacterium tuberculosis lineage
Association between sociodemographic/clinical factors and M. tuberculosis drug resistance
Association between Haarlem or non-Haarlem M. tuberculosis strains and sociodemographic or clinical factors
Relative growth of M. tuberculosis lineage 7 strains.Cultivation of the M. tuberculosis lineage 7 isolates on Middlebrook 7H10 agar plates showed significantly smaller colony diameter and weight among lineage 7 strains compared to those in lineages 3 and 4 combined (Fig. 4).
Growth of Mycobacterium tuberculosis lineage 7 versus other lineages on agar plates. M. tuberculosis isolates belonging to lineages 3, 4, and 7 were streaked onto Middlebrook 7H10 plates and incubated at 37°C for 5 weeks, and colony morphology parameters were studied. (A) Colony diameter. The diameters of 4 to 8 representative colonies were measured, and average values were calculated for each isolate. The data presented are mean diameters ± standard deviations (SD) among the isolates. (B) Colony weight. Three representative colonies were picked up, and dry weight was measured for each isolate. The data presented are mean colony weight ± SD among the isolates (*, P < 0.000).
DISCUSSION
This study documents a very high genetic diversity of modern and premodern M. tuberculosis lineages among PTB patients in Amhara Region, Ethiopia. Our previous study provided preliminary evidence that a high proportion of PTB patients in this geographic region carry strains belonging to lineage 7 (12). M. tuberculosis lineages, including lineage 7 strains, are along with other lineages thought to have emerged approximately 70,000 years ago, near the time modern humans migrated out of Africa (2).
Criteria for identifying lineage 7 strains were previously established (9, 20–22) and were adhered to in the present study. Accordingly, all lineage 7 strains identified here (n = 36) have deletions from the 4th to 24th spacers and have a high degree of sequence similarity to the genomes of previously reported lineage 7 strains (9). The presence of two or more repeats at the 24th MIRU-VNTR locus is also considered highly discriminatory for modern and ancient M. tuberculosis lineages, as reported in genotyping studies from Bagdad (20), Singapore (21), and Bangladesh (22). All 36 strains in the present study have two alleles at the 24th MIRU-VNTR locus. Moreover, phylogenetic tree analyses support this result; both UPGMA and MST analyses clearly show that the 36 strains reside in a single branch separate from other M. tuberculosis lineages. These data confirm the spoligotyping results of our previous study, in which we concluded that the 36 SIT 910 and 1729 strains belong to lineage 7 (12).
Notably, we observed longer patient delay in seeking treatment associated with infection with lineage 7 strains. A possible explanation is that infection with lineage 7 strains may more often cause mild rather than severe symptoms, and that these patients wait longer before seeking medical attention. This may result in silent TB cases, increasing the risk of transmission to uninfected individuals. Thus, more intense efforts to identify active cases are recommended in order to control the spread of TB disease. The duration of TB symptoms, time of health care seeking, and associated factors have been studied in many countries (23–29). Various sociodemographic and clinical factors have been identified as predictors of delay in seeking care. However, to our knowledge, an association between M. tuberculosis lineage and patient delay has not been reported previously. Our results suggest for the first time a possible relationship between M. tuberculosis lineage-specific disease and health-care-seeking behavior. The longer patent delay in seeking treatment observed in the M. tuberculosis lineage 7 strain might be explained by the longer time required for cultivation of these particular M. tuberculosis strains.
Another implication from this study is that M. tuberculosis strains belonging to lineage 7 may be less virulent than strains of other M. tuberculosis lineages. Our study demonstrated that the average diameter of colonies and colony weight were significantly smaller for lineage 7 strains than those for strains of other lineages studied. These observations indicate that lineage 7 strains grow more slowly than non-lineage 7 M. tuberculosis strains in vitro. However, it is not known whether M. tuberculosis lineage 7 strains also grow relatively slowly in vivo, nor is it known whether slow growth of lineage 7 strains is associated with low virulence.
A previous study showed that Beijing strains of M. tuberculosis are more likely to cause severe disease and to develop multidrug resistance than other strains. It has been proposed that the virulence of these strains is due to their ability to produce a phenolic glycolipid (PGL) that reduces the efficiency of the host immune system (4). Although many M. tuberculosis strains do not produce PGLs, it is not yet known whether lineage 7 strains do or do not produce a similar compound. Further proteomic and lipidomic studies are warranted to address this question.
Clustering is a marker of recent transmission (30, 31). We observed a clustering rate of 32%, which is lower than that in a similar study that showed 45.1% clustering in Ethiopia (8). A recent TB transmission index in a study population can be used to assess the efficacy of the TB control program in a certain geographical area (30). The relatively lower clustering rate observed in our study may be linked to the general trend of decreasing TB incidence and prevalence in Ethiopia. The health extension program is functioning well at the community level, and suspected TB patients are generally detected and referred for diagnosis and treatment within a few weeks of onset of symptoms. However, the clustering rate observed in our data indicates an unacceptable rate of transmission. It is desirable to reduce the rate of transmission, and this will require increased effort to actively trace patients' contacts within the suspected transmission area. Haarlem strains were more likely to cluster, and these strains were more likely to demonstrate resistance to streptomycin and any of the first-line anti-TB drugs compared to other lineages. A similar finding was previously observed in Ethiopia (8). This suggests high rates of transmission of drug-resistant M. tuberculosis Haarlem strains. The Amhara Region has nearly 20 million inhabitants, and there is only one TB culture facility in the region. It is crucial to expand the capabilities for TB culture and drug susceptibility testing to combat the spread of drug-resistant TB in the region. Patients carrying NW-ETH3 strains were also more likely to cluster, and these strains were more likely to be resistant to any of the first-line anti-TB drugs, as observed in another study from Ethiopia (8).
Among the modern M. tuberculosis lineages identified in our study, the majority, 60 (26.4%), belonged to CAS/Delhi. Another study from Ethiopia reported 38.9% CAS strains (8), and CAS/Delhi was the predominant lineage in a study from Sudan (32). CAS/Delhi strains have been circulating in Central Asia, the Middle East, Saudi Arabia (33) and India (34) for a long time. The presence of a high proportion of CAS/Delhi strains in this study may be due to the fact that CAS/Delhi predominates in this region, or the presence of these strains may be related to the increase in migration, trade, and tourism between Ethiopia and the Middle East.
The novel M. tuberculosis lineages identified in this study were designated NW-ETH1, NW-ETH2, NW-ETH3, and NW-ETH4. According to the SIVIT/SPOLGD4 database, all strains in these groups belong to the ill-defined T spoligotype. Strains with the name NW-ETH3 are similar to spoligotype T3-ETH (http://www.pasteur-guadeloupe.fr/tb/spoldb4/spoldb4.pdf). NW-ETH3 strains were more likely to cluster and be resistant to any of the first line anti-TB drugs. A previous study demonstrated a high frequency of clustered cases of infection with novel multidrug-resistant T3-ETH strains of M. tuberculosis in Ethiopia (35).
In summary, our study demonstrates that the TB epidemic in the study area is caused by a highly heterogeneous group of M. tuberculosis strains belonging to modern and premodern M. tuberculosis lineages. Patients with lineage 7 strains had increased duration of TB symptoms and were more likely to delay seeking medical attention. Haarlem strains frequently demonstrated resistance to one or more first-line anti-TB drugs. The considerable clustering of M. tuberculosis cases indicates ongoing transmission. Increased efforts to identify active cases and trace patient contacts will be essential to contain the spread of disease. Additional laboratory facilities in which TB culture and drug susceptibility testing can be performed are needed in the study region.
ACKNOWLEDGMENTS
We are grateful to Norwegian Institute of Public Health, Oslo University Hospital, and the Amhara Regional State Health Bureau for facilitating this study. We thank all of the study participants for providing valuable samples and the health care workers for facilitating sample and data collection.
This work was supported by the Research Council of Norway, GLOBVAC Program (grants 192468/S50 and 234506/H10 to G.N. and T.T., respectively).
FOOTNOTES
- Received 19 December 2014.
- Accepted 29 January 2015.
- Accepted manuscript posted online 11 February 2015.
- Copyright © 2015, American Society for Microbiology. All Rights Reserved.
