Spatial clustering and genetic diversity of Mycobacterium tuberculosis isolate among pulmonary tuberculosis suspected patients, Arsi Zone, Ethiopia

The study was conducted at the Asella Teaching Hospital, University hospital of Arsi University, situated in Asella Town, South-East of Addis Ababa. The hospital has 297 beds and acts as a medical referral center for a population of 3.5 million inhabitants in the Arsi Zone and its surroundings. The Zone is found in the central part of the Oromiya National Regional State and astronomically lies between 60 45′ N to 58′ N and 38 32′ E to 40 50′ E. Currently, it is divided into 25 districts: Amigna, Aseko, Bale Gasegar, Chole, Digeluna Tijo, Diksis, Dodota, Enkelo Wabe, Gololcha, Guna, Hitosa, Jeju, Limuna Bilbilo, Lude Hitosa, Merti, Munesa, Robe, Seru, Sire, Sherka, Sude, Tena, Tiyo, Ziway Dugda. Provided that Asella Teaching Hospitals is one of the oldest health facilities with better infrastructure and experienced health personnel as well as the only referral hospital in the Zone, most tuberculosis patients particularly referral cases come to the hospital from all districts.

A facility-based study was conducted at Asella Teaching Hospital between February 2016 and August 2016. Before the sputum collection, a pretested structured interviewer-administered questionnaire was used to collect data on socio-demographic, (age, sex, residence, education), districts of the patients (≥ 7 years old), previous contact with TB cases, and facility service-related data from pulmonary tuberculosis suspected patients followed by the collection of sputum samples at a spot and the next day morning for sputum smear examinations. Age of ≥ 7 years old was considered since they could provide sputum sample after consent was obtained from the parent and/or legal guardian.

As depicted in Fig. 1, sputum smear positive samples were collected and mycobacterial culture was conducted as described by Forbes et al. [13]. Briefly, the samples were decontaminated and concentrated using N-acetyl L-cysteine (NANC)–NaOH techniques at Adama regional tuberculosis laboratory. The concentrated samples were suspended in 1–2 ml phosphate buffer solution (pH 6.8), thoroughly mixed and 2–3 drops were inoculated into the Lowenstein Jensen (LJ) media and incubated for 4–8 weeks with a regular check-up for the growth. Colonies from culture-positive samples were preserved in 10% glycerol freezing media until required for further processing. Isolation of genomic DNA (gDNA) was conducted on culture-positive isolates using the Roche Cobas Amplicor extraction kit (Roche Diagnostics, USA) as previously described [14]. The DNA samples were shipped to the Laboratory of the Health, Technology and Informatics, The Hong Kong Polytechnic University for genotyping.

The PCR amplification of the 24 MIRU-VNTRs loci was conducted using 24 pairs of primers as described previously by Supply et al. [15]. Briefly, amplification was conducted in 96 well plates that enabled the amplification of eight samples per 12 loci. The target loci were amplified in a total of 20 µl mix containing variable proportion of MgCl₂, 4 µl of Q-solution, 0.08 µl (5 U/µl final concentration) HotStarTaq DNA polymerase, 0.4 µl (10 mM) dNTP, 0.4 µl forward and reverse primers (at 20 µM final concentration), 2 µl 10X PCR buffer, 9.1 µl, and 2 µl of DNA template. The amplification was conducted using Veriti thermal cycler set at initial denaturation at 95 °C for 15 min, 40 cycles of 94 °C for 1 min, annealing at 59 °C for 1 min, extension at 72 °C for 1:30 min, and a final extension of 72 °C for 10 min. The amplified products were mixed with 6X loading dye in a proportion of 5:2 v/v, electrophoresed for 2 h using 1.5% agarose gel (Vivantis Agarose (Molecular Biology Grade) in 1X TBE (Tris–Borate EDTA) buffer with RedSafe™ Nucleic Acid Staining Solution (iNtRON Biotechnology) and visualized under UV transillumination. The product size was compared against the 100 bp ladder and the copy number was determined manually using allele calling tables. Each locus for eight isolates was electrophoresed together to reduces the bias during size determination (Additional file 1: Figure S1).

After the collected data were entered into the excel spreadsheet, it was exported to SPSS version 22 for descriptive and Binary logistic regression model analysis. For spatial analysis, using the retrospective discrete bernoulli model, the data was exported to SaTScan version 9.6 and Quantum Geographical information system (QGIS) version 3.12. The binary logistic regression model was used to determine the association of patients’ characteristics with sputum smear positivity, while the retrospective discrete Bernoulli model was used to identify the sputum smear positivity hot spot districts. The maximum spatial cluster size was considered as 50% of the population at risk; the Boscoe’s limit on cluster risk level (relative risk) of more than one was used to restrict the high rate cluster. The final significant clusters were declared at a p value less than 0.05 (i.e. which combined the Standard Monte Carlo, Sequential Monte Carlo, and Gumbel Approximation) with its 95% CI of relative risks and those which had no Geographical Overlap. The maximum replication was set at 999 of Monte Carlo Replication. Finally, sputum smear-positive hot spot areas were presented by QGIS as per their level of cluster. An analysis of the MIRU-VNTR data was conducted using the browser-based software (http://​www.​MIRU-VNTRplus.​org) as described by Allix-Beguec et al. [16]. The numerical 24 loci copy numbers of isolates were entered into an excel spreadsheet and uploaded. Similarity matching with the standard reference strains was conducted by setting the maximum distance of 0.17 followed by phylogenetic tree classification. Unweighted Pair Group Method with Arithmetic Mean (UPGMA) based dendrogram of the phylogenetic tree was constructed. A cluster rate was calculated as the number of isolates with identical MIRU-VNTR alleles divided by the total number of test isolates. The minimum spanning tree was used to detect the clonal complex of M. tuberculosis as well as to assess the genetic links and clonal complexes (CC). The clonal complex was defined as M. tuberculosis strains with only two loci differences.

A total of 832 PTB suspected patients participated in the study. Out of this, 451 (54.2%) were male and 589 (70.8%) were rural residents. The median age of the patients was 35 years with an interquartile range of 27 and age ranged from 7 up to 86 years. A total of 324 (38.9%) of the patients were illiterate and the majority, 658 (79.1%) had no previous history of hospital admission (Table 1).

Out of 832 presumptive PTB patients, 119 (14.3%) were smear-positive with variable tuberculosis prevalence among the studied sociodemographic characteristic. Accordingly, male (15.1%), age group of less than or equal to 25 (22.7%), and rural residents (15.8%) had a higher prevalence of smear-positive PTB. Out of the variables considered in this study, only age groups stand out to be significantly associated with the smear-positive rate (X² = 28.26, P = 0.000). Further analysis of the considered variables with the prevalence of smear-positive PTB showed that the age and residence of the patient were the risk factors for the smear positivity. Accordingly, people at age groups of 7–25 and 25–34 were 4.53 (AOR = 4.53; 95% CI 2.25–9.13) and 3.00 (AOR = 3.00; 95% CI 1.41–6.35) times at higher risk of turning smear-positive compared with the old age group (≥ 55 years), respectively. Likewise, though the residence site of the patients was not identified as a risk factor on the bivariate analysis, it becomes a risk factor after adjusting for other confounders on the multivariate analysis (Table 2).

As illustrated in Fig. 2, presumptive PTB patients visited the hospital from 23 (92%) of the currently 25 districts of Arsi Zone. The disease was significantly clustered in eleven (11) (47.8%) districts (RR = 2.27; 95% CI 1.62–3.2) of the zone while two districts (i.e. Digaluna Tijo, and Tiyo) were identified as potential clusters of the disease (RR = 1.43; 95%CI 0.95–2.16) (Table 3).

Out of 90 culture-positive isolates, seventy-two (72 (90%)) isolates were genotypically characterized using standard MIRU-VNTR. Of 72 isolates queried for the lineage assignment, 59 (81.9%) were classified into the previously known lineages and 13 (18.1%) were not assigned to any known lineages. Overall, 42 (58.3%) belong to M. tuberculosis lineage 4 (Euro-American), 16 (22.2%) M. tuberculosis lineage 3 (Delhi/CAS), and 1 (1.4%) M. tuberculosis Lineage 1 (Indo-Oceanic/East Africa Indian). Further classification to the sublineage indicates that the predominant lineage was Delhi/CAS comprising 16 (22.2%) isolates followed by 15 (20.8%) isolates belonging to Haarlem. The remaining isolates were distributed as 13 (18.1%) TUR, 6 (8.3%) LAM, 4 (5.5%) URAL, 4 (4.5%) NEW-1 and 1 (1.4%) EAI (Table 4). The sub-lineages were distributed across the districts (Additional file 1: Table S1). Further, the comparative results of the study isolate and isolates in the MIRU-VNTR database are provided in Additional file 2: Figure S2. A table comprising information of MIRU-VNTR characterized isolates with country of origin, distrcits of each patient, name and order of each locus is provided as Additional file 3: Table S2 (xls). The constructed phylogenetic tree exhibited a distinct clustering by lineages and sub-lineages as well as isolates were highly diverse and had deep branching (Fig. 3). The analysis of the clustering rate indicated 17 isolates out of 72 (23.6.0%) were grouped into 4 clusters with each cluster composed of 2–7 isolates belonging to the Haarlem sub-lineage. Delhi/CAS and LAM sub-lineages consisting of 1 cluster each composed of 2 isolates.

The minimum spanning tree (MST) was constructed to identify the clonal complex groups of the isolates within dual locus variants. As shown in Fig. 4, the lineages are indicated by a different color and, the size of the individual circle shows the number of clusters. All the lineages detected using UPGMA analysis were also detected in MST that further confirm genetic linkage and clustering. Six clonal complexes (CC1-6) consisting of a maximum of 6 and a minimum of 2 isolates with dual locus distance were identified. The largest clonal complex (CC1) (5 isolates) belongs to the TUR sub-lineage.