Background
The
Mycobacterium tuberculosis complex (MTBC) constitutes a group of mycobacteria which are 99.9% similar at the nucleotide level and the causative agents for tuberculosis (TB) [
11]. Globally, TB became the leading cause of death from an infectious disease [
39]. Ethiopia stands 12th in the world and 4th in Africa among the high TB burden countries with 24,000 TB deaths and 165,000 new TB cases in 2018 [
39]. The current prevalence of MDR/RR-TB in Ethiopia is 0.71 and 16% in new and previously treated TB cases, respectively [
39].
Understanding the molecular epidemiology of TB is important for regional disease control. For instance, distinct strains may be linked to outbreaks [
10], high virulence [
42], emergency of drug resistance [
44], disease progression [
43], and can point to the geographic origin of a strain [
20,
33] as well as identify new lineages [
17,
28].
The South Omo Zone is an administrative unit in the southern Ethiopia bordering Kenya and South Sudan. The area is remote with a poor infrastructure and high population diversity with 16 different ethnic groups. Forty-two percent of South Omo’s residents including 15 ethnicities have a pastoral life style. The facilities for health care and education are underdeveloped especially in the pastoral regions [
4]. A previous study suggested that the prevalence of TB among pastoralists is higher than in other socio-economic groups in Ethiopia [
24]. In depth, high resolution molecular epidemiological surveys are required to characterize the diversity of MTBC isolates in this remote pastoral region precisely.
Beginning in the 1990s a number of molecular genotyping techniques have evolved to differentiate MTBC at the species and strain levels [
22]. Whole genome sequencing is ideal to identify a strain type, but technically and informatically demanding and too expensive to characterize regional MTBC diversity [
22]. Widely used in TB research are spacer oligotyping (spoligotyping) and mycobacterial interspersed repetitive units – variable numbers of tandem repeat (MIRU-VNTR). Spoligotyping targets a single locus and has less discriminatory power but is simple and cost effective. MIRU-VNTR targets numerous loci with increased discriminatory power. International databases and data analysis tools were created for both genotyping methods [
22].
Most MTBC genotyping studies in Ethiopia employed spoligotyping [
3,
6,
7,
16,
19,
26,
43]. We are aware of only very few Ethiopian studies using spoligotyping and MIRU-VNTR simultaneously to profile MTBC strains [
1,
9,
36,
37,
43]. New lineages such as the MTBC lineage lineage_7/Aethiops vertus [
17,
28], Ethiopia_2 and Ethiopia_3 [
37] were newly assigned. Most surveys were in geographic areas more accessible than South Omo. The objective of the current study is to examine the MTBC population structures in the latter region and compare the data to those of other Ethiopian regions and globally.
Discussion
This study was the first of its kind to analyze the MTBC population structure and transmission dynamics in the South Omo Zone, southern Ethiopia. The study included PTB and EPTB patients, and identified highly diverse lineages. The clustering rate/RTI was low in the study area. Logistic regression analysis showed that clustering of strains was associated with SIT status.
Health facilities other than JGH in the study area are in range of 14 to 250 km from Jinka town where JGH and the Regional Laboratory are located. Due to feasibility, samples were stored in health facilities at -20 °C from a week to 3 weeks. The variation in culture recovery rate of MTBC isolates in this study possibly associated with sample storage conditions. There was continuous electric interruption in the Zone which affects the storage temperature which could compromise the viability of MTBC in the sample. JGH had its own backup generator that might be the reason for better recovery rate for samples from JGH. In connection, the overall culture recovery rate in this study is less than previous studies in Ethiopia [
47,
48].
In contrast to previous reports in the study area [
8,
41], the number of EPTB cases in this study was low. From personal observations, the low number of EPTB cases in this study might be due to lack of skilled pathologist to take FNA samples whereas in previous studies relied on clinical symptoms. Spoligotyping and MIRU-VNTR are recommended methods for the profiling of MTBC isolates [
23,
35]. Both genotyping methods in this study were in range of highly discriminant [
27,
34]. MIRU-VNTR has higher discriminatory value than spoligotyping as shown here and in earlier studies [
1,
9,
36,
37].
Spoligotyping of South Omo MTBC isolates resulted in a clustering rate of 57.7%%. This rate agrees with a previous study in Gambella, Southwest Ethiopia [
3] which is geographically proximate to the present study site. The rate is lower than a national survey [
19] and that of studies in Addis Ababa [
26], Northwest Ethiopia [
37], Eastern Ethiopia [
7], and central Ethiopia [
6]. But it was a study in Western Ethiopia [
16]. The MIRU-VNTR clustering rate was 3.9%. The rate is lower than other studies in Ethiopia [
1,
9,
36,
37,
43] and higher than a Chinese study [
15]. Such variability in clustering rate among studies could be due to differences in geography, population density, ethnicity and socio-economic diversity [
31]. The low clustering rate in our study could also be associated with low culture recovery rate of samples which make potential isolates from the study population not to be genotyped and/or presence of low TB transmission in South Omo due to geographic expanse which disfavor the transmission as a result of very less crowdedness in the community.
Most MIRU-VNTR alleles in this study were highly and moderately discriminant based on the allelic diversity (
h) which is an indirect indicator of the sample representativeness of the study population [
34]. Values of
h more than 0.8 and less than 0.1 are unsuitable for genotyping [
30]. In the present study, all MIRU-VNTR loci were suitable for genotyping of the isolates except locus 960 with an
h value of 0.810.
This study identified six major lineages (EA, EAI, IO, lineage_7,
M. bovis and
M. africanum). Four isolates were identified using TB insight database as
M. africanum which is known to be localized in West Africa. However, three of the four isolates were re-identified using updated version of SITVITWEB as Ethiopian and considered as Lineage_7/Ethiopia_1 [
9,
17,
37] but one isolate remained unknown. The population in the area is endogenous which make the plausibility of an unknown isolate to be
M. africanum less probable. In general, this might imply the need of updating TB Insight database with newly generated MTBC data from the horn of Africa, particularly Ethiopia.
EA is the most dominant lineage in the world [
32], ranging in Ethiopia from 32.5% near the border to South Sudan [
3] to 86.8% in central Ethiopia [
6]. This might highlight the introduction of EA lineages from abroad through the capital city and their expansion to the peripheral areas. In addition, the existence of EA in high number in contrast to Ethiopian lineage, Lineage_7, in the country might indicate the high transmission ability of EA. The contribution of
M. bovis for TB was low which is supported by data from other studies [
7,
17]. While larger clinical studies are needed, our data suggested that the role of
M. bovis as a causative agent of TB in pastoral area presumably linked to contact with infected cattle and, consumption of raw milk and meat [
5].
According to the updated version of SITVITWEB database, the T lineage was predominant in this study. This lineage accommodates MTBC strains which do not have phylogeographic specificity [
12]. Among the T sub-lineages, T3 (42.2%) was the most dominant one in this study. It is also called Ethiopia_2 [
9] which is followed by T3-ETH, also called Ethiopia_3 [
9]. These isolates supposed to be phylogeographically specific to Ethiopia including well defined Ethiopian lineage also called lineage_7 [
17]. The CAS1_Delhi lineage was the second predominant lineage in the study area which was followed by Haarlem. The Haarlem is believed to descend from the European continent [
12]. The predominance of the T lineage in Ethiopia and the CAS lineage in Tanzania [
25] and Kenya [
29] supports the notion of enrichment of MTBC strains in certain geographies. The Turkey lineage was present in the study area which is believed to be specific to Turkey [
45]. This might be associated with presence of Turkey investors in the South Omo. In addition, Ural_1, Manu, Bov, EAI_SOM, and X lineages were identified in less frequency in the area. It is plausible that the observed phylogeographic diversity has linked to considerable international tourism in South Omo, Ethiopia.
When we look at MIRU-VNTR
plus based lineage in the present study, Delhi/CAS was the predominat one which is in agreement with previous studies in Ethiopia [
9,
36,
37,
43]. Ethiopia_2 is the predominant Ethiopia specific lineage followed by the Ethiopia_3 and lineage_7, similar to a previous study in geographic proximity, Southwestern Ethiopia [
36]. But in studies at far distance, in Northwestern Ethiopia [
9,
43], the predominant lineage was lineage_7 followed by the Ethiopia_3 and Ethiopia_2 lineages. These findings indicate that the distribution of Ethiopia specific lineages differ moderately from area to area within the country localities. This information is useful for the country’s TB Control Program. Almost 6% of isolates in this study were not assigned into lineages which requires further study and introduction into the genotype database. The relationship among lineages in the MST based on MIRU-VNTR loci was in agreement with similar studies conducted in Ethiopia [
9,
37].
Based on the generated data from spoligotyping and MIRU-VNTR, it is possible to say that these two methods can complement each other. But they have different precisions. For instance, MIRU-VNTRplus can identify 37 Delhi/CAS, 22 Haarlem and 22 Ural lineages whereas SITVITWEB can only identify 17 CAS1_Delhi, 16 Haarlem and eight Ural_1. This differences probably associated with the algorithm used by such databases. Finally, from all isolates of South Omo in this study, MIRU-VNTRplus and SITVITWEB didn’t assign nine and 56 isolates into lineages, respectively.
The multivariate logistic regression analysis in this study showed none of the variables had association with strain clustering except SIT shared status. Orphan strains were less likely to cluster compared to shared strains which implies that shared strains have higher transmission rate compared to orphan strains in the study area.
We contend that the number of genotyped isolates is sufficient for a primary representation of South Omo’s MTBC population structure, assessment of clustering rates and RTIs. However, having less culture recovery rate from samples other than JGH in this study limited our ability to identify clusters and RTIs more comprehensively.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.