Drug resistance and population structure of M.tuberculosis isolates from prisons and communities in Ethiopia

Despite recent achievements seen in the fight against tuberculosis (TB), it still remains a significant cause of morbidity and mortality in Ethiopia [1]. This situation is worsened by an increase in prevalence of multidrug-resistant (MDR) Mycobacterium tuberculosis complex (MTBC) strains, defined as resistance to at least the two most powerful first line anti-tuberculosis drugs; isoniazid (INH) and rifampicin (RIF) [2].

The recent TB drug resistance survey, conducted in Ethiopia from 2011 to 2013, revealed an increase in MDR-TB with a prevalence of 2.3% and 17.8% among new and previously treated cases, respectively [3]. These data indicate that MDR-TB has been becoming a significant public health threat in the country. The occurrence and transmission of MDR-TB in confined environment like correctional facilities could further worsen the problem [4]. Due to the specific conditions found in prisons such as crowded living conditions, a large number of inmates could be infected and develop active TB disease within a short period of time. Accordingly, TB incidence and MDR rates in prisons have been found to be higher compared to that reported in the general population in several studies [5‐7]. This is likely to have an impact also on the civilian population e.g., by direct transmission via visits, prison staff or later spread in community after release from prison [7, 8]. Importantly, latently infected inmates may become a reservoir of MDR-TB and a threat to the communities. In some studies performed to identify the risk factors for TB in communities previous incarceration history was described as a risk factor [9, 10].

While a number of studies confirm that TB rates are higher in prisons [7, 8, 11] little direct evidence about strain diversity and population structure of M.tuberculosis within prisons and the interaction of the prison setting with the community e.g., by molecular epidemiological studies is available [4, 12].

Increasing evidence suggests that the underlying genetic diversity of the MTBC has a significant impact on the pathogenicity and immunogenicity of individual strains, thus, knowledge of the regional population structure linked with phenotypic data such as with the drug resistance pattern could be relevant for the implementation of an effective TB control program tailored to specific genotypes and local circumstances [13]. Further, molecular epidemiological studies have been instrumental to define recent transmission dynamics in various settings as well as to describe the local and global population structure of the MTBC [13]. Modern molecular DNA fingerprinting methods like Mycobacterial Interspersed Repetitive Unit-Variable Number Tandem Repeats (MIRU-VNTRs) typing and spacer oligonucleotide typing (spoligotyping) are highly suited to investigate both, the population structure and transmission of the MTBC in communities or special settings such as prisons [14].

To date, no molecular epidemiological studies have been carried out to decipher MTBC strain diversity in prisons in Ethiopia. Only few studies investigated the population structure of MTBC strains in communities of central and northwestern parts of the country by spoligotyping. Due to the limited discriminatory power of this method, however, these studies reported the “ill defined” T linage as predominant strain [15‐17]. There are two studies in which the combined methods of MIRU-VNTR and spoligotyping were applied, and in which Delhi/CAS was the dominating MTBC lineage [18, 19]. These two studies included only the north-western part of Ethiopia.

In order to contribute to a more comprehensive knowledge on MTBC strain diversity in the whole of Ethiopia, we conducted a pilot study to determine the population structure and drug resistance pattern of MTBC strains isolated from prisons and communities of southwestern, southern and eastern Ethiopia by combined application of MIRU-VNTR and spoligotyping. We further investigated to which extent the isolated strains were related to each other by the calculation of clustering rates.

A total of 127 MTBC strains were initially isolated from sputum specimens collected in prisons (24) and hospitals (103). Out of these, 18 (14.17%) isolates were excluded from the final analysis: ten strains could not be reactivated in (liquid) culture, five isolates were mixed infections and three isolates had repeatedly inconsistent DST results.

The basic information on the 109 participants and strains included in the final analysis are depicted in Table 1. As expected, there was a significant gender difference between participants from prisons, where 90.48% were male, and hospitals with 51.14% males (p < 0.001). The mean age of the study subject was 29.03 (95% CI; 26.89, 31.2) with no significant difference between prison inmates and hospital participants (t = 0.93). All 88 participants from hospitals and 17 out of 21 (80.95%) inmates from prison had never received previous TB therapy. MTBC strains from prison inmates were collected in Oromia and Southern Nations, Nationalities and Peoples Regional State (SNNPRS) regional states. TB isolates from hospital patients were additionally collected in Somali regional state and Dire Dawa city administration.

In this study, we analyzed the MTBC strain population in prisons and communities of southern, southwest and eastern Ethiopia using combined spoligotyping and MIRU-VNTR typing methods. In agreement with results of previous reports from Ethiopia we could also show a high strain diversity in our study [19, 25, 32]. About 34% of the MTBC strains analyzed are Ethiopian specific Lineages and sub-lineages, Lineage 7, Ethiopia_H37Rv like and Ethiopia_3, which were described recently and have not yet been reported elsewhere [18, 19, 33]. Opposed to other studies conducted in northwest Ethiopia Delhi/CAS was not the dominating lineage [18, 19]. Tessema et al. hypothesized that the influx of Indian and Chinese peoples to Ethiopia due to growing business relations with Ethiopia introduced the Delhi/CAS lineage in the country [18]. If this hypothesis was true, the Delhi/CAS lineage dominancy might have started from the center Addis Ababa and is now processing to the periphery. This could explain why in some remote areas investigated in this study, where a relevant proportion of the population are still leading a nomadic life style which is driven by the search for water and grazing land for their cattle, not Delhi/CAS but the Ethiopian lineages are still dominating.

Clustering is a marker of recent transmission [34, 35] and knowing the clustering rate of TB strains that circulate in the community can help to evaluate the performance of TB control programs or to formulate new control strategies. The overall clustering rate in this study was 31.19% which was lower than the previously reported 45.1% from Ethiopia [18]. On the other hand our clustering rate is consistent with the 32% reported for Amhara regional state study [19]. This data could suggest that a relevant proportion of active TB cases were due to reactivation of latent infection. Indeed, there was a significant decline in TB prevalence observed in Ethiopia in the past five years [36]. This decline in active TB cases might have contributed to the lower clustering rate seen in this study compared to older reports. The cluster analysis of our strains collected in 2013 and strains analyzed by Tessema et al. in 2009 revealed several clusters, including strains from both studies, indicating that some strains remain in the population and lead either to reactivation of a remote infection or are effectively transmitted over a four years timeframe in Ethiopia [18]. Interestingly, transmission rates of active TB seemed to be higher in Somali regional state. Strains from this regional state showed a significantly higher clustering rate compared to other regional states. This finding could be explained with the geographic context of Somali regional state which is bounded by Djibouti, Somalia and Kenya [37] where a free and intensive movement of peoples living in the border areas might have contributed to an ongoing TB transmission.

As there was no statistical difference between the clustering rates in the communities (31.82%) and prisons (28.57%), our data suggest that transmission rates are not higher in prisons than in communities. This comparison, however, is hampered by the slightly different study regions, the low number of MTBC strains isolated from prison inmates and the different approaches and coverage of MTBC strain collection in prisons and communities. On the other hand we could show that MTBC strains which were isolated in prisons build clusters with strains collected from community members. This finding could indicate that TB infection was acquired in the community and developed later to active TB disease under the specific conditions found in prisons. This hypothesis is further supported by findings which were previously published by our group [7]. In that specific study, a contact with a TB case at home before incarceration was one of the strongest risk factors for active TB disease in prison [7].

Despite the great achievements seen in Ethiopia to reduce overall TB prevalence [1], the control of MDR-TB still seems a distant prospect. In this study we have observed an MDR-TB prevalence in the community of 2.27% which was consistent with the 2.3% of the Ethiopian public health institute (EPHI) survey report performed in year 2014 [3]. However, comparing our findings with 0.8% seen in 2002 [38] and the 1.6% estimation of WHO in 2011 indicates that MDR-TB is increasing with alarming rate through time [3, 39]. In prisons, the MDR TB prevalence (9.52%) and also the number of subjects with TB in the past was higher than in communities, although, this difference was not statistically significant in our analysis of a limited number of isolates. This result is alarming since prisons may act as a reservoir for MDR-TB in the country. The occurrence of a single MDR-TB case in prison might have huge implications for prison health and the community, considering bidirectional communication between both.

This study has several limitations and the findings should be interpreted with care. First, due to a relative short study period, the number of MTBC isolates in this study, specifically the number of strains from prisoners, is lower than in other reports [18, 19]. Therefore, the magnitude of clustering and also specific risk factors on clustering, e.g., drug resistance, could have been undetected or underestimated. Second, only TB strains from selected communities (which had access to a TB laboratory located in neighboring hospitals) and prisons (to which the investigators had permission to enter) were included in the analysis. This fact could introduce selection bias, especially as only characteristics of TB strains from participants who had access to the health system or who were inmates in the rather huge zonal prisons could be studied. Third, clustering rates could be imprecise and rather reflecting the spread of dominant strains types but not recent transmission as not all MTBC strains in the study areas were analyzed but only those collected in the catchment area of large hospitals and large prisons. Fourth, the rather cross-sectional study approach in both prison and community settings did not allow for the investigation of risk factors of TB transmission in cases with an incubation period of more than 12 or five months, respectively. Fifth, HIV test results were not available for community based patients. This hampered the analysis of the influence of HIV on TB transmission and clustering. Finally, this pilot study highlights the need to further investigate the drug resistance, population structure and transmission dynamics of TB in communities and prisons in Ethiopia as well as the interaction of both groups in a larger and prospective survey.

Our study provides first data on MTBC population structure and drug resistance pattern of strains found in Ethiopian prisons and in regional states of the country which were not studied before. Our findings suggest that TB is still not sufficiently controlled in specific, potentially remote, areas of the country and highlight the need for improved tools and new strategies aiming for MDR-TB control, especially in prisons. The fact that TB strains from prisoners are forming clusters with community based TB strains is worth noting and stresses the importance of the inclusion of prisons in strategies for TB control in the whole of Ethiopia. Future studies of sufficient duration and area-wide strain collection need to be performed to improve our knowledge on risk factors for TB transmission in Ethiopia.