Background
Tuberculosis (TB), caused by
Mycobacterium tuberculosis complex
(M. tuberculosis), remains one of the deadliest infectious diseases worldwide. In 2014, 9.6 million people contracted TB and 1.5 million died from the disease [
1]. The number of deaths due to TB slowly declined between 2000 and 2013 due to effective diagnosis and treatment, but remains unacceptably high [
2].
Spoligotyping is a method investigating the diversity at a highly variable CRISPR locus evolving by deletion in
M. tuberculosis complex [
3]. Spoligotyping has been widely used to classify
M. tuberculosis clinical isolates by family and subfamily [
4,
5] that were later found to be largely concordant with lineages as defined by an whole genome sequencing (WGS) approach, allowing single nucleotide polymorphisms (SNPs) to be identified [
6]. In addition and despite a relatively low discrimination level of spoligotyping [
7,
8] it can be used for first genetic identification of patient’s clinical isolates and suggest or exclude recent transmission cases which should be investigated further using more discriminatory methods [
7,
9,
10]. Previous studies demonstrated that spatial clustering of TB data when associated to genetic clustering of TB cases more easily allows to focus on adequate settings to distinguish most vulnerable populations and reactivation versus recent transmission cases [
10‐
13].
Hence, geospatial tools may be helpful to study the TB dynamics of urban areas with high prevalence of TB. Geospatial tools incorporating Geographic Information Systems (GIS) enable the identification and mapping of spatiotemporal clustering of disease or patients [
14]. The GIS method has been used to study the spatial distribution of human TB cases and has identified the heterogeneity of epidemic areas [
10,
15]. To confirm recent TB transmission, isolates must be found to be clonal. Clonality must then be investigated by more discriminatory genotyping methods. In Antananarivo, an exhaustive thorough genomic characterization of clinical isolates either by 24 MIRU-VNTR or by WGS remains out of reach for economic reasons for the time-being. For this reason, we chose a classical spoligotyping approach, which remains a first-line method to characterize clinical isolates in resource-limited countries. Such a combination of approaches (spatial and genetic clustering) is interesting to locate spatial clusters of TB and attempt to assess where recent TB transmission cases may occur.
In Madagascar, the incidence of TB in 2013 was estimated to be approximately 233/100,000 inhabitants [
16]. TB prevalence distribution in Madagascar is likely heterogeneous with particularly high rates in specific areas driven by uncontrolled transmission, as in most resource-limited countries with either remote or isolated settings [
15,
17,
18]. A crucial element of TB control efforts is the identification of “TB Hotspot areas” for the orientation of TB control strategies given the lack of resources. Therefore it would be beneficial for the TB program to have a tool for targeting areas of high transmission risk where interventions should be concentrated. Previous studies using genotyping techniques on clinical isolates have shown a large diversity of circulating
M. tuberculosis genotypes in Madagascar [
19‐
21]. A preliminary study based on TB notification rates identified spatial aggregation of TB cases in Antananarivo [
15,
17]. This aggregation could be due either to actual transmission and/or to reactivation cases. Our aim here was to accumulate evidence concerning the possibility that previously identified TB hotspot areas in Antananarivo could be linked to transmission events.
We used a combination of spatial tools and genotyping to identify potential high-prevalence and likely higher recent transmission risk, TB areas in Antananarivo, Madagascar, as done recently in Brazil and Japan [
11,
12]. These methods will be used in Madagascar for further understanding of the TB epidemiology in Antananarivo and the identification of priority targets for TB control strategies.
Discussion
This study aimed to determine the spatial signature of TB by using a combination of genotyping and geospatial tools across the urban city of Antananarivo, Madagascar. The complementarity of both the genotyping and spatial analyses approaches has been used for the determination of TB transmission areas and risk studies in Brazil and Japan [
11,
12]. These two methods are not new but they had never been combined for the detection of potential high risk areas of TB transmission in Antananarivo.
The combination of GIS and spoligotype data identified four hotspots with potential TB transmission in the city of Antananarivo. Analysis of all TB cases (without distinction of genotypic clustering of strains) identified a TB disease focal point overlapping with one of the potential transmission area (constituted by the Fokontany of Andohatapenaka II). TB hotspots such as that detected by the analysis taking only spatial data into account is likely constituted simultaneously by patients linked by indirect and relatively ancient TB transmission and by patients linked by local and recent TB transmission. Spatial clustering of patients with genotypically clustered isolates may therefore concern more patients associated with local and recent transmission. Additionally, our combination of approaches permitted the detection of spatial clusters of TB patients which were not detected with only the spatial scan of TB cases. TB transmissions might have occurred in these areas. For definitive proof of TB transmission, more discriminatory genotyping tools should however be used.
While previous studies determined risk factors associated with spatial clustering of TB cases in Antananarivo [
15,
18], risk factors associated with transmission were poorly investigated. The first potential transmission hotspot is constituted by the Fokontany of Antohomadinika Afovoany that is localized in one of the poorest neighborhood in the 1st urban district of Antananarivo. Most of the houses in this area are made of precarious wooden hovels and the majority of the local population does not have standard health care access. While life under fragile conditions and environmental factors [
14,
19] are known factors that contribute to TB reactivation [
16], they could also foster local transmission as suggested by this study. The relatively high diversity of SITs seen in this poor area supports the previous findings that environmental factors also favor reactivation of latent TB. This area, containing the highest rate of TB, constitutes the first high risk area of TB transmission in Antananarivo.
The three other areas of potential TB transmission (#2, #3 and #4) host or lie close to three public markets (Anosibe, Isotry and Ambanidia market) where there is an important flow of people. Similar studies have shown that a flow of persons promotes TB transmission [
11,
12]. This study further adds evidence for potential transmission in markets.
The discriminatory level of spoligotyping is relatively low, and the genotypic clustering cannot be used to estimate TB recent transmission rates. Thus isolates with the same spoligotype may be coming from different chains of transmission. However, when compared with isolates with single genotype, that might represent reactivation or relatively far recent transmission, these clustered isolates are more likely to have originated from recent spatially localized transmission. The aim of the study was to identify high risk areas, this limitation does therefore not invalidate this study and should be taken as a first step towards identification of real TB transmission hotspots.
From 1994 until 2000, studies of the
M. tuberculosis genotype profiles present in Antananarivo reported six lineages and sub-lineages of
M. tuberculosis clinical isolates: T, LAM, and H (Lineage 4), EAI (Lineage 1), CAS (Lineage 3), and Beijing (Lineage 2) [
19]. The 6 lineages were shown to be present in the capital of Madagascar in 2004/2005, with the appearance of other minor sub-lineages such as S and X along with some unknowns (U) [
27]. The distribution of the major circulating clinical isolates observed in these studies may have evolved slightly, but we noted no significant change since then, suggesting that all the lineages were being continuously transmitted in this area.
In this study, the population density in every studied Fokontany was taken into account, increasing the accuracy of spatial clustering identification. The use of spoligotyping as a first screen is a relatively simple and inexpensive method. It allowed us to identify clusters that were overlooked using only spatial information. We thus plan to keep this strategy as a first-line detection of potential transmission areas. A possible improvement in our approach would be the use of methods providing both spoligotyping and resistance data such as TB-SPRINT [
28]. This high-throughput assay tests for rifampicin and isoniazid resistance simultaneously with spoligotyping on Luminex. It could be a useful strategy in the near future to survey and prevent the spread of MDR-TB cases.
Another limit of this study was the short recruitment time (9 months). However, this recruitment period was sufficient to achieve a large sample size, and given the stability of the population, such duration should not have created too much bias. Some of the enrolled patients did not agree to give consent to participate in the study, those residing in Antananarivo Renivohitra consulting with DTCs outside Antananarivo, and those not consulting with a DTC had to be removed from the inclusion. Finally, we chose the patient residence to perform the spatial analysis although the patient residence is clearly not the unique site of possible TB transmission. Activity areas have been linked to transmission in other studies [
11,
12]. We still chose to locate TB cases according to the residential Fokontany since people with most poor living condition and without work circulate most of their time in the same Fokontany in Antananarivo. A further investigation using the combined spatial methods on both working and residential areas could still be useful for TB epidemiological surveillance.
Acknowledgements
The “Projet interne” of the Institut Pasteur de Madagascar funded the work. Our thanks go to the “Programme National de Lutte Contre la Tuberculose” represented by the Dr. Andrianantenaina RAKOTOSON for the authorization to conduct this study, the officials and staff of the various “Centre de Diagnostic et de Traitement de la Tuberculose”, the Mycobacteria Unit of the IPM, the SIG cell of the Epidemiology Unit of IPM, and the IGEPE team of the Institute of Integrative Cell Biology, UMR9198, CEA-CNRS-University Paris Saclay, France, for their collaboration. We are also grateful to the TB patients who accepted to participate and provided biological samples to perform this study and the Development Office of Antananarivo (DOA) for the different datas used for the study.