Use of genotyping based clustering to quantify recent tuberculosis transmission in Guadeloupe during a seven years period: analysis of risk factors and access to health care

Background The present study aimed to characterize Mycobacterium tuberculosis population structure and to identify transmission chains and risk factors by prospective molecular typing in conjunction with conventional epidemiological investigations in the French overseas department of Guadeloupe. Methods The study included all the culture-positive TB cases (1 clinical isolate per patient; n = 129) diagnosed between a seven year period (April 4th, 1999 to December 31st, 2005). Prospective molecular typing was performed using spoligotyping and VNTRs, and a subset of 44 M. tuberculosis isolates found to be clustered was retrospectively typed using 12-loci MIRUs. Data were compared using the SITVIT2 database, followed by analysis of risk factors in function of clustering of the isolates and available demographic and socioeconomic data. Results The study sample was characterized by a majority of new cases (87.4%); a moderate proportion of drug-resistance (7.8%); a high level of immigration (51.2% foreign-born) originating from high TB/HIV incidence neighboring islands such as Haiti or Dominican Republic; lower socioeconomic conditions (70.7% of jobless, average income 824 EUR/month); and a significantly higher proportion of TB/HIV co-infected cases (38.2% vs. 8.5%; p < 0.001), and extrapulmonary disease (18.2% vs. 4.8%; p < 0.02) among migrants as compared to French patients. The study revealed an important delay in access to healthcare with a median delay of 74.5 days between the 1st symptoms and clinical suspicion of TB. Prospective molecular typing based on spoligotyping and 5-loci VNTRs showed that evolutionary recent Euro-American lineages predominated in Guadeloupe (91.5% of isolates). In conjunction with epidemiological data, it allowed to estimate a recent transmission rate of 18.6%, which was close to the rate of 16.7% estimated using retrospective 12-loci MIRU typing. Although a higher proportion of cases in older age-group were apparently linked to reactivation; univariate analysis of risk factors did not allow pinpointing specific risk factors for a patient to belong to a TB transmission group. Conclusions Ongoing TB transmission in the insular, low TB-incidence setting of Guadeloupe can be defined as follows: (i) a significant proportion of imported cases of the disease from neighboring islands; (ii) significantly higher TB/HIV coinfection among foreign-born cases; and, (iii) a higher proportion of cases affecting older age-group among French patients due to reactivation. This study emphasizes the need for universal typing using spoligotyping and 15-loci MIRUs in prospective studies.


Background
There has been major progress in reducing tuberculosis (TB) cases in the past two decades, yet the global burden of TB remains enormous. The World Health Organization (WHO) estimated that there were 8.7 million incident cases of TB and 1.4 million deaths in 2011 [1]; HIV pandemic apparently played a major role in it, since 13% of the TB patients were co-infected with HIV and accounted for 30.7% of all deaths due to TB. Even if most of the TB cases occurred in Asia (59%) and Africa (26%), a non negligible proportion was reported in the Eastern Mediterranean (7%) and European (5%) regions, followed by the Americas (3%). To achieve the current goal of a high level of case detection (75%) and cure (85%) by the year 2015, a finer study of the epidemiological characteristics of TB patients to pinpoint risk factors involved and access to health care is needed in different regions of the world [2]. Molecular fingerprinting of Mycobacterium tuberculosis has been proposed as an ideal investigational strategy to achieve this objective, since it allows to incorporate genotyping based clustering to quantify recent TB transmission in order to analyze risk factors and access to health care [3][4][5]. We hereby report results of such an investigational approach under a prospective genotyping study covering TB transmission during a 7 year period (1999)(2000)(2001)(2002)(2003)(2004)(2005) in Guadeloupe, a French overseas department of the Americas. Under this study, all the M. tuberculosis strains were characterized by spoligotyping [6] and Variable Number of Tandem DNA Repeats (VNTRs; [7]) that were methods of choice when this investigation was planned back in 1998, with the aim of determining potential transmission links between TB patients. This prospective study also undertook concomitant investigations to study the access to health care and follow-up of the TB patients diagnosed.

Study design, patients, and M. tuberculosis isolates
After a prior explanation of the recommended investigations for the diagnostic purposes by a physician, patients were asked to sign an agreement letter and the pathological specimens were obtained from suspected tuberculosis patients. Only patients residing in Guadeloupe for more than three months were eligible. The cultures were performed using Löwenstein-Jensen slants at 37°C, and were duly identified as M. tuberculosis complex using classical biochemical tests and the AccuProbe test (GenProbe Inc., San Diego, CA). All positive cultures were also subjected to drug-susceptibility testing using the proportional method on Löwenstein-Jensen media. The study performed between April 4 th , 1999 and December 31 st , 2005 corresponded to a recruitment of 129 TB patients with one M. tuberculosis clinical isolate per patient. For each patient, the data requested included nature of pathological specimen, result of AFB smear, the date of diagnosis, the clinical form of the disease, results of HIV serology, and previous history of TB. Demographic and socioeconomic data as well as information relating to the diagnosis were collected by the investigator after individually interviewing each patient using a specific form.

Ethical considerations
The study protocol was presented to the Conseil National de l'Ordre des Médecins (http://www.conseilnational.medecin.fr/qu-est-ce-que-l-ordre-1206), which guarantees the quality of healthcare provided to the patients. As recommended, it was thereafter submitted to the Commission nationale de l'informatique et des libertés (CNIL; http://www.cnil.fr/english/the-cnil/operation/#c1556) which ensures the protection of personal data. Patients were duly informed about the description and aims of the study in writing, and those enrolled after their oral consent were asked to sign an agreement letter, returned to investigators. The study was approved by CNIL under a formal agreement number 999343 (modalities of registration, storage and analysis of epidemiological, clinical and socioeconomic data are available online from: www.pasteur-guadeloupe.fr/tb/projects/tuberculose.pdf ).

Molecular typing
Right from the beginning of the prospective study in 1999, the molecular typing was planned using two PCRbased systems -spoligotyping [6] and VNTRs [7], since this typing scheme is significantly cheaper and faster than the historical "gold standard" IS6110-RFLP that was used for earlier M. tuberculosis genotyping work in Guadeloupe [8][9][10]; yet provided with results that were relevant for epidemiological investigations around clustered cases of TB [11]. Briefly, the genomic bacterial DNA was prepared by the cetyltrimethylammonium bromide (CTAB) method [12] from cultured strains. Spoligotyping using primers designated DRa and DRb (with DRa biotinylated at 5') to amplify the whole Direct Repeat (DR) region was performed as described previously [6]. As spoligotyping used alone may overestimate the number of potentially linked isolates, it was used in association with VNTR typing using exact tandem repeats ETR-A to ETR-E as a 2 nd -line VNTR typing method as reported [7].
After the completion of the study, a subset of 44 M. tuberculosis isolates found to be clustered by combination of spoligotyping and 5-loci VNTRs was retrospectively typed by using the 12-loci MIRU-VNTRs system described later [13]; since it allows a better discrimination than the 5-loci VNTRs that were available when the prospective study was planned back in 1998. The PCR-based 12-loci MIRU-VNTR typing was performed using primers described previously [13,14], and the presence and size of each PCR product was determined manually by electrophoresis on an agarose gel, followed by staining with ethidium bromide as reported previously [15]. Note that although theoretically ETR-D and E correspond to MIRU locus 4 and 31, the primers and calculations based on ETR determination according to the protocols of Frothingham and Meeker-O'Connell [7] versus Supply et al. [13,14], might not turn up in identical copy number for ETR-D and MIRU-4.
Genotype analysis, database comparison, and recent transmission rate Spoligotype patterns as octal codes, 5-loci VNTRs, and 12-loci MIRU-VNTR data were entered in the SITVIT2 proprietary database of the Institut Pasteur de la Guadeloupe, which is an updated in-house version of the publicly released SpolDB4 [16] and SITVITWEB [17] databases. In this database, Spoligotype International Type (SIT), VNTR International Type (VIT), and MIRU International Type (MIT) designate identical patterns shared by 2 or more patient isolates, whereas "orphan" designates patterns reported for a single isolate that does not correspond to any of the patterns recorded in the repository of the SITVIT2 database. Major phylogenetic clades were assigned according to signatures initially provided in SpolDB4, and slightly revised in SITVITWEB by the addition of 5 "new rules" for definition of variants within 62 existing lineages/sub-lineages. These include specific signatures for various M. tuberculosis complex members, as well as rules defining major lineages/sub-lineages for M. tuberculosis stricto sensu, i.e., the Beijing clade, the Central Asian (CAS) clade and two sublineages, the East African-Indian (EAI) clade and nine sublineages, the Haarlem (H) clade and three sublineages, the Latin American-Mediterranean (LAM) clade and 12 sublineages, the "Manu" family and three sublineages, the S clade, the IS6110-low-banding X clade and four sublineages, and an ill-defined T clade with five sublineages. Lastly, the recent transmission rate was estimated by the N-1 method [5], according to formula T(c) -N(c)/T(a), where T(c) is the total number of clustered isolates, N(c) is the number of clusters, and T (a) is the total number of isolates.

Epidemiological, phylogenetical, and statistical analysis
Epidemiological and demographic data collected were analyzed using Epi-Info 3.2.2 software (Centers for Disease Control and Prevention, Atlanta, GA; free download available at http://www.cdc.gov/epiinfo). The statistical analysis was carried out with STATA® 9.0 software (StataCorp, College Station, TX). Patients were divided into 2 groups: clustered patients (infected by strains harboring identical genotypic patterns), versus unclustered patients. The Chi-2, Fischer exact, and Student tests (according to the variables considered) were carried out using univariate analysis in order to test risk factors in function with clustering of the isolates. Odds ratio were calculated with a confidence interval of 95%. A qualitative analysis of the data collected was performed to evaluate the possible ways of infection. The phylogenetical analysis of genotyping data was carried out using "PAUP 4.0" software (available at: http:// paup.csit.fsu.edu) and UPGMA (Unweighted Pair Group Method with Arithmetic Averages) algorithm on combined genotyping data.

Results
Demographic characteristics, geographic localization, and drug resistance Epidemiological and demographic data of the 129 patients with a culture-positive, confirmed M. tuberculosis infection are summarized in Table 1. The population of TB patients in Guadeloupe is characterized by a male to female sex-ratio of 1.5 with a median age of patients of 40 years. Overall TB principally affected men in the age group 40-59 years (n = 32/77 or 41.6% of all male patients); male patients being older than females (median age of 42 vs. 36 years). More than 1 out of 2 patients was foreign-born (63/129 or 48.8% French vs. 66/129 or 51.2% foreign-born), who had been residing in Guadeloupe for approximately 10 years at the time of diagnostic. The exact place of birth was available for 126/129 patients: 53/126 (42.1%) were born in Haiti, 5/ 126 (4%) in Dominican Republic and 52/126 (41.3%) in French overseas departments (Guadeloupe n = 43, Martinique n = 1, French Guiana n = 1), or other departments in continental France (n = 7). Foreign-born patients were significantly younger (median age 36 years vs. 50 years for French patients; p < 0.001); the proportion of patients below 60 years being 63/66 or 95.5% among foreign-born vs. 40/63 or 63.5% for French patients (p < 0.001). Lastly, men were older than females among the foreign-born group: 39 vs. 34 years as compared to 49 vs. 58 years for French nationals (difference not significant statistically). These results corroborate published data from mainland France for French vs. foreign-born patients for the year 2004 [18]. Geographic localization of patients showed that 27/34 cities of Guadeloupe archipelago (including its island dependencies of Saint-Martin, Saint-Barthelemy, Marie-Galante, Desirade, and Les Saintes) declared at least one case of tuberculosis during the period of the study; 6 areas shared 69/116 (59.5%) of the TB cases for which this information was available: Saint-Martin 20/116 (17.2%); Les Abymes 20/116 (17.2%); Le Gosier, 10/116 (8.6%);

Access to healthcare and delay to diagnosis
Access to healthcare was defined as the delay comprised between the first symptoms of TB and the date of its clinical suspicion. This interval includes 2 periods: the access time to 1 st medical contact which is the delay comprised between the 1 st symptoms of TB and the date of 1 st medical contact for those symptoms, and the time of clinical suspicion defined as the delay extending from the date of 1 st medical contact to the date of clinical suspicion of TB. In the present study, the median access time to 1 st medical contact and to clinical suspicion were equal to 16 and 26.5 days respectively leading to a median delay to healthcare of 74.5 days. We observed that patients treated after clinical suspicion presented a significantly longer median delay of access to healthcare than those treated after bacteriological investigations (median delay of 80 vs. 37 days respectively; p < 0.017); due to a longer access time to 1 st medical contact (median delay of 19 vs. 8 days respectively) and a significantly longer delay of clinical suspicion (33.5 vs. 10.5 days respectively; p < 0.03). Remarkably, in both groups of patients, the diagnostic delay was longer than the access time to the first medical contact. The primary reason for consulting a physician for a first medical contact was persistent cough with fever (median 16 days). TB was suspected at the 1 st medical contact for 69/107 (64.5%) patients, 2 nd contact for 22/107 (20.6%), patients, 3 rd contact for 10/107 (9.3%) patients, and later for 6/107 (5.6%) patients. Patients for whom a presumptive diagnosis of TB was made at the 1st medical contact had a longer delay in consulting the physician after the onset of clinical symptoms (30.5 vs. 3.5 days for other patients; p < 0.0001). Nonetheless, these were diagnosed for TB significantly faster than the remaining patients (10.5 vs. 74 days; p = 0.025) most probably due to a more advanced state of the disease; which suggests that efforts targeting early medical diagnosis of TB must be intensified.

Socioeconomic characteristics
As summarized in Table 1, only 27/92 (29.3%) patients had a regular work (employees n = 13, workmen n = 4, farmers n = 4, craftsmen n = 3, management n = 2, and unspecified job n = 1).  (Table 1). Among those lost of sight (n = 33), 26 patients followed their initial treatment for an unspecified period of time (median delay of 4 months after the initial phase of treatment in hospital, no information available for 7 patients).
We also looked in the updated version of the SITVIT2 database for the geographical distribution of all clustered isolates found in Guadeloupe vs. their relative distribution in all the 3 French Departments of the Americas (Guadeloupe, Martinique, and French Guiana), Caribbean, South and central America, as well as the remaining countries worldwide (database interrogation made on April 25 th 2013; Table 2). One can observe essentially 5 kinds of situations: (i) ubiquitous spoligotypes present in a very high proportion in the database (SIT42/LAM9, SIT50/H3, SIT53/T1), (ii) patterns that are shared between the 3 French departments, Caribbean, South and Central America and the rest of the world, although in varying proportions (SIT2/H2, SIT3/H3, SIT7/T1, SIT20/ LAM1, SIT44/T5, SIT45/H1, SIT51/T1, SIT93/LAM5, SIT122/T1), (iii) patterns with a phylogeographical specificity for South and Central America that are present in Caribbean including Guadeloupe (SIT17/LAM2); (iv) patterns that are shared between Caribbean and the rest of the world but absent in South and Central America (SIT5/T1, SIT70/X3); (v) patterns that show phylogeographical specificity for French Departments in general and Guadeloupe in particular (SIT14/X3, SIT103/H1, SIT1087/T1).
Genotyping of the 5-loci VNTRs provided complete and/or interpretable results for 116/129 strains (Figure 1) with 15 clusters (n = 106 isolates, 2-27 strains/cluster). Note that 9 of the 10 unclustered strains (with the exception of pattern 96245) matched an existing VIT number in the database, and no VIT numbers were newly created due to a newly found pattern or a match with preexisting orphan in the database.
The combination of spoligotypes and VNTRs revealed 76 distinct profiles for the 129 isolates: 58 corresponded to unique vs. 18 for clustered strains (n = 71, 2-9 isolates/cluster; labeled as A, B1, B2, C1, C2, C3, D, E, F, G1, G2, G3, G4, H, I1, I2, I3, and J in Figure 1). When epidemiological data were analyzed in combination with clustering, we observed that mean age of patients among the 18 clusters varied from 27 (cluster I1, n = 2) to 59 years old (cluster G2, n = 3), and that the male to female sex ratio varied from 0 (cluster E, n = 2) to 3 (cluster A, n = 4); with 4 clusters composed exclusively of It should be underlined that the cluster I1 composed of the youngest patients, did not contain any drug resistant strains, and that the 3 patients infected with MDR strains were simultaneously HIV-positive. A comparison with the SITVIT2 database showed that the combined spoligotyping + 5-loci VNTR profiles of the 6 clusters including resistant strains, matched mainly strains from Haiti. For example, 9/10 strains in the database sharing cluster J pattern (SIT5/VIT193) were isolated from Haitian patients. Similarly, 14/16 patients with cluster B2 pattern (SIT93/VIT3); 20/24 strains with cluster C2 pattern (SIT17/VIT10), 26/36 strains with cluster C3 pattern (SIT20/VIT10), and 34/84 of strains with cluster G4 pattern (SIT50/VIT18), were also isolated from Haitian patients in the database. Unfortunately, drug resistance profiles were not systematically available in the database, making it impossible to definitely link the genotypes and drug resistance of isolates actively circulating between Haiti and Guadeloupe.

Retrospective evaluation of clustered isolates by 12-loci MIRU typing
Because during the time course of this 7 years prospective study, the TB molecular typing methodology evolved with the introduction of 12-loci MIRU typing [13,14], we decided to retrospectively genotype a subset of 44 M. tuberculosis strains clustered by the combination of spoligotyping and 5-loci VNTRs. The results obtained (Table 3) showed that 26/44 (59.1%) strains remained clustered vs. 18/44 (40.9%) being unique by the combination of the 3 methods. This allowed decreasing the clustering rate and recent transmission rates in this subset from 55% and 41.1% respectively, to 25.5% and 16.7% respectively.

Contact tracing investigations and risk factor analysis
Rebuilding the ways of infection of TB cases due to a recent infection was carried out by cross analyzing genotyping data in conjunction with results of contact tracing investigations, as well as data collected through individual interviews. Contact tracing alone suggested a potential common source of infection for 6 patients of the study, which in our opinion is underestimated since only 54% of the TB cases diagnosed could be effectively covered by investigations among patient's entourage. Genotyping results combined with epidemiologic investigations underlined 23 TB cases categorized as recent transmission. Finally, the consideration of all the 3 parameters confirmed active transmission of TB for 24/129 (18.6%) of cases.
A Univariate analysis of risk factors was performed in function to the clustering of M. tuberculosis isolates and available demographic, epidemiologic, clinical, and socioeconomic data. However, the results summarized in Table 4 showed that none of the risk factors tested could be significantly associated with active transmission of TB. Only the factor "age of patients below 60 years" was close to the statistical limit of significance (p = 0.06).

Discussion
This study is the first prospective epidemiological and molecular study of TB transmission in the French overseas department of Guadeloupe. When compared to TB epidemiology in mainland France, the most striking difference was the significantly higher proportion of TB/ HIV co-infected patients in Guadeloupe (24.5% vs. 10% respectively; [19]). However, an almost similar proportion of drug-resistance and TB incidence were observed in 2005: the proportion of drug resistant isolates being 7.8% vs. 8.8%, respectively, in Guadeloupe and continental France [19], and an incidence of 8.1 vs. 8.9 cases per 100,000 inhabitants [20]. Similarly to mainland France, men represented around 60% of TB patients in Guadeloupe a trend accentuated for the age group 40-60 years in mainland France (68% of TB patients) and 40-59 years in Guadeloupe (71.1% of TB patients [20]). Lastly, TB in Guadeloupe was characterized by immigration with 51.2% of foreign-born patients in our study, which is slightly higher than 48% in mainland France in 2005, but significantly lower than the 69% of foreign born patients observed in French Guyana (185/268; p < 0.001; [20,21]). Immigrants in Guadeloupe mainly originated high TB incidence and AIDS burden surrounding Caribbean islands of Haiti and Dominican Republic, with reported TB incidence in 2005 of 305 and (See figure on previous page.) Figure 1 UPGMA dendrogram obtained through a combined numerical analysis of spoligotypes and VNTR profiles of the 129 M. tuberculosis clinical isolates from Guadeloupe. The letters in front of the tree correspond to the following information: A: SIT numbers of the spoligotype profiles with, in brackets, the strain numbers; B: binary spoligotypes; C: 5-loci VNTR profiles. The "x" symbol among a profile indicates a missing or uninterpretable result; D: clusters numbers; E: clades attribution of the spoligotype profiles.  91 cases/100 000 inhabitants [22], which might explain the relatively high rate of TB/HIV coinfected patients among foreign born patients in our study (38.2% vs. 8.5% for French patients; p = 0.001).
The present study was meant to specifically evaluate the use of genotyping to quantify recent TB transmission with a focus on risk factors and access to health care during 1999-2005; and was followed by a study on evolution of TB cases during 2006-2011 [23]. We therefore attempted to compare TB evolution during these 2 time periods. As summarized in Table 5, one may notice a statistically significant decrease during the period 2006-2011 for 4 out of 9 parameters compared, namely, the proportion of foreign-born patients; patients lost of site; excessive alcohol users; drug addicts. The decrease in the proportion of foreign-born patients from 51.2% vs. 38.3% may be linked to the decrease in the proportion of migrants observed in Guadeloupe from 6.5% to 4.7% in 2010 [24,25]. It may be noted that the decrease among lost of site is statistically significant both among foreignborn and French patients, while the decrease in excessive alcohol users and drug addicts was limited to French patients only. Regarding the 5 parameters that did not change significantly (Table 5), it might be worthwhile to comment on trends for TB/HIV coinfection which remained stable between the periods 1999-2005 and 2006-2011 (24.5% to 26.6%), despite a slight but non-significant increase among both French-born (8.5% to 13.9%) and foreign born patients (38.2% vs. 46.9%). However, considering the reported decrease in the incidence of HIV/AIDS infection in Guadeloupe from 20 to 11.7 cases/100 000 inhabitant between 2003 and 2009 [26,27], this observation might be explained on the basis of continued influx of a varying proportion of migrant workers from poor surrounding countries such as Haiti and Dominican Republic, where both TB and HIV/AIDS  (Table 5). In our study, foreignborn patients were more likely to be TB/HIV co-infected and also showed a higher proportion of extra pulmonary disease. These observations are likely not independent, although stratified analysis was not performed to be conclusive.
Regarding geographical mapping of TB cases, 5 out of 6 cities sharing 60% of TB burden in our study were characterized by higher population densities than the mean reported for Guadeloupe, and 3 out of 6 were also characterized by a higher proportion of underprivileged populations (more than 12.6% [28,29]). Furthermore, TB patients in our study were characterized by a very high rate of unemployment (70.7%; Table 1) and a below average income close to the official limit of poverty in France (monthly income of 824 EUR vs. 812 EUR/month in 2001; [29,30]).
Our results also provided with a first insight into the management of TB in Guadeloupe, and revealed that 38/107 (35.5%) of patients for which information was available, consulted more than one physician before being diagnosed for TB. Interestingly, the management of patients improved in the period 2006-2011 [23] with a median delay of 55 days  between the 1 st symptoms and clinical suspicion of TB instead of 74.5 days in 1999-2005; as well as the delay for clinical suspicion of TB after a 1 st medical consultation (18 days vs. 26.5 days). Comparison of foreign-born vs. French patients did not reveal significant differences in access to healthcare. Regarding treatment outcome, only 55.8% of patients were cured in our study since 15.0% of them deceased, and 29.2% were lost to follow up. These relatively mediocre TB management indicators as compared to Europe (74% treatment success rate, 7% deceased, 12% lost to follow up; [31]), may indirectly be linked to the lack of awareness among the patient population in Guadeloupe, which is characterized by poor living conditions underlined earlier.
As far as feasibility of molecular typing in a prospective study is concerned, our results demonstrated that PCRbased spoligotyping not only correctly identified potential "outbreak" strains, but also allowed a fine characterization of the phylogeographical specificity of certain clones, e.g., SIT14/X3, SIT103/H1, and SIT1087/T1 ( Table 2). The 2 ndline genotyping using 5-loci VNTRs allowed reducing the overestimation of clustering by spoligotyping [32]. However, it was apparently not enough discriminatory since 12-loci MIRUs performed retrospectively allowed to reduce both the clustering and recent transmission rates (Table 3). This observation argues for the use of spoligotyping with extended MIRU typing formats. In our study, mixed clusters involving both foreign-born and French patients concerned 11/18 or 61% of clusters by spoligotyping and 5-loci VNTRs; however after retrospective 12-loci MIRU typing their number was reduced to 5 mixed clusters containing a total of 18 cases. Thus the suggested minimal requirement based on our study is 12-loci MIRUs, although the optimal format for molecular epidemiological investigations today is a new 15loci format [33,34]. Nonetheless, seeing the very low recent transmission rates found in our study, the future investigations should be extended to the sister island of Martinique, as well as French Guianasince important migratory flows and exchanges exist between Guadeloupe and the 2 other French departments of the Americas. This might help detect both migratory flows, as well as direct and casual contacts, leading to the ongoing TB epidemic.
Last but not least, thanks to the comparison with an updated SITVIT2 database, we were able to conclude on the phylogeographical specificity of the circulating M. tuberculosis clones in Guadeloupe and surrounding regions. Most of this data commented earlier in the text clearly shows the predominance of evolutionary recent Euro-American group of isolates [16,17] in our study. However, LAM, Haarlem and T lineages predominated in Guadeloupe, while the X lineage strains of suspected Anglo-Saxon descent [16,17,35] were present to a smaller extent. Thus the bulk of TB transmission in Guadeloupe may be attributed to TB lineages shared with neighboring areas in the Caribbean and South and Central America [21,36]. Nonetheless, we further identified some clones such as SIT14/X3, SIT103/H1, and SIT1087/T1 (Table 2), which are highly specific for Guadeloupe. We now hope to extend a detailed molecular epidemiological study on the M. tuberculosis isolates from the 3 French departments isolated in 2006-2012 so as to see the evolution of above mentioned genotypes in time and space. Ultimate comparison of clusters retained after 12and 15-loci MIRUs should be done using 24-loci MIRU-VNTRs [34] followed by the whole genome sequencing [37] of specific clones, in order to shed light on epidemiological and phylogenetical links between strains involved in active transmission, as well as to quantify how much M. tuberculosis isolates vary at the genomic level between epidemiologically linked patient clusters [38].

Conclusions
The present prospective genotyping study aimed to investigate TB transmission during a 7 year period in Guadeloupe, as well as to study the access to health care and follow-up of the TB patients diagnosed. The results obtained underline that the ongoing TB transmission in this insular, low TB-incidence setting of Guadeloupe was characterized by: (i) a significant proportion of imported cases of the disease from neighboring islands; (ii) a significantly higher TB/ HIV coinfection among foreign-born cases; and, (iii) a higher proportion of cases affecting older age-group among French patients due to reactivation. Last but not least, this study also emphasized the need for universal typing using spoligotyping and 15-loci MIRUs in future prospective studies in Guadeloupe.