|Home | About | Journals | Submit | Contact Us | Français|
Genetic tracking of Mycobacterium tuberculosis is a cornerstone of tuberculosis (TB) control programs. The RDRio M. tuberculosis sublineage was previously associated with TB in Brazil. We investigated 3847 M. tuberculosis isolates and registry data from New York City (NYC) (2001–2005) to: 1) affirm the position of RDRio strains within the M. tuberculosis phylogenetic structure, 2) determine its prevalence, and 3) define transmission, demographic, and clinical characteristics associated with RDRio TB.
Isolates classified as RDRio or non-RDRio M. tuberculosis by multiplex PCR were further classified as clustered (≥2 isolates) or unique based primarily upon IS6110-RFLP patterns and lineage-specific cluster proportions were calculated. The secondary case rate of RDRio was compared with other prevalent M. tuberculosis lineages. Genotype data were merged with the data from the NYC TB Registry to assess demographic and clinical characteristics.
RDRio strains were found to: 1) be restricted to the Latin American-Mediterranean family, 2) cause approximately 8% of TB cases in NYC, and 3) be associated with heightened transmission as shown by: i) a higher cluster proportion compared to other prevalent lineages, ii) a higher secondary case rate, and iii) cases in children. Furthermore, RDRio strains were significantly associated with US-born Black or Hispanic race, birth in Latin American and Caribbean countries, and isoniazid resistance.
The RDRio genotype is a single M. tuberculosis strain population that is emerging in NYC. The findings suggest that expanded RDRio case and exposure identification could be of benefit due to its association with heightened transmission.
Mycobacterium tuberculosis is a major human pathogen that has undergone clonal evolution into divergent lineages that are associated with specific geographic regions and possibly with distinct human ethnic populations (Hirsh et al., 2004; Gagneux et al., 2006). The major M. tuberculosis lineages and select sublineages can be defined by specific chromosomal large-sequence polymorphisms (Tsolaki et al., 2004; Reed et al., 2009) and/or groups of single nucleotide polymorphisms (SNPs) (Gutacker et al., 2002). Lineages may also be estimated through molecular epidemiologic tools, including IS6110-restriction fragment length polymorphism (RFLP) analysis (van Embden et al., 1993), spoligotyping (Brudey et al., 2006), and mycobacterial interspersed repetitive units (MIRU)-typing (Supply et al., 2000).
Recent reports have suggested that genetic variation between M. tuberculosis lineages may translate into measurable biological differences. Lineage-specific differences in rate of growth, ability to induce or evade immune responses, pathogenicity, and expression of virulence factors have been observed previously in vitro (Manca et al., 2004; Pheiffer et al., 2005; Riley, 2006) and/or in animal studies (Lpez et al., 2003; Reed et al., 2004; Tsenova et al., 2005). Lineage-specific differences in transmissibility and the resulting clinical manifestations of human tuberculosis (TB) have been previously suggested (van Crevel et al., 2001; Sun et al., 2006; Hanekom et al., 2007a; Kong et al., 2007; de Jong et al., 2008; Feng et al., 2008; Thwaites et al., 2008; van der Spuy et al., 2009). More data are needed to clarify the potential differential capacity of lineages to transmit and cause disease which, in turn, could influence local TB disease burden (Gagneux et al., 2006) and impact public health strategies for TB control, such as the prioritization of certain strains for investigation. Even though, M. tuberculosis lineage-specific differences in host ethnic preference are beginning to emerge (Gagneux et al., 2006; Caws et al., 2008), confirmation by large and longitudinal data of combined heterogenous human and M. tuberculosis populations is needed.
The Latin American-Mediterranean (LAM) family of M. tuberculosis is responsible for ~15% of TB cases reported in the spoligotype-defined SpolDB4 database (over 39,000 international strains) and, as such, is a major M. tuberculosis lineage (Brudey et al., 2006; Lazzarini et al., 2007). We recently described the RDRio genotype as a clonally-derived sublineage within the LAM family (Lazzarini et al., 2007). It is defined by a ~26.3 kb deletion resulting in the loss or modification of 10 genes, including two PPE mycobacterial family genes known to be recognized by host immunity (Singh et al., 2005; Gibson et al., 2008). RDRio M. tuberculosis strains (hereafter called RDRio) possess a coincident deletion, RD174, at a separate genomic locus (Gibson et al., 2008), and that classifies it as part of the Euro-American/African superlineage (Reed et al., 2009). RDRio is the most prevalent cause of TB in Rio de Janeiro, Brazil (Lazzarini et al., 2007), and is present in Europe, Africa, and the Americas (Gagneux et al., 2006; de Jong et al., 2008; Gibson et al., 2008). It has been associated with a more “clinically extensive” form of TB with increased sputum bacillary load, lung cavities, hemoptysis, and weight loss, suggesting potential for increased transmissibility (Lazzarini et al., 2007; Lazzarini et al., 2008). It was therefore important to ascertain if the potentially heightened transmissible nature of RDRio is maintained outside of endemic areas, particularly in locales with a high diversity of human ethnic and racial groups and multiple circulating M. tuberculosis lineages; New York City (NYC) is just such a context.
TB in NYC (Li et al., 2003), like the rest of the United States (Talbot et al., 2000), is increasingly a disease of foreign-born individuals. Foreign-born TB cases are often infected by lineages prevalent in their native region (Hirsh et al., 2004). Knowledge of the lineages circulating in NYC and their relative capacity to spread within the multiethnic environment of NYC is therefore critical to understanding current and future M. tuberculosis transmission trends. This study aimed to: 1) affirm the phylogenetic position of the RDRio sublineage within the global M. tuberculosis population structure, 2) determine the prevalence of RDRio, and 3) define transmission, demographic, and clinical characteristics associated with RDRio TB in comparison with other lineages of M. tuberculosis.
From 2001–2005, 5508 cases of TB were diagnosed in NYC; 4202 (76.3%) were culture-positive for M. tuberculosis. IS6110-RFLP and spoligotyping data, as well as DNA for further analysis, was available for 3911 (93%) isolates. Isolates from cases with two different strains (n = 22) and those with an indeterminate RDRio/non-RDRio genotype (n = 42) were excluded from the analysis, leaving a study total of 3847 TB cases. The incidence of TB in NYC in 2005 was 12.3 per 105 (among non-US born 24 per 105(TB Annual Summary, 2005). IRB approval for the study was granted by the Weill Medical College of Cornell University and the NYC Department of Health and Mental Hygiene (NYCDOHMH).
M. tuberculosis isolates from NYC TB cases routinely undergo IS6110-RFLP, spoligotyping, and MIRU-typing (Clark et al., 2006). In addition, SNP cluster analysis was performed on all isolates from 2001 to 2004 (Gibson et al., 2008), while isolates from 2005 were assigned a SNP cluster based on IS6110-RFLP and spoligotyping data by investigators at the Public Health Research Institute. Spoligotype family was assigned using SpolClust [http://cgi2.cs.rpi.edu/~bennek/SPOTCLUST.html] with consideration of IS6110-RFLP pattern and major SNP-based cluster data (Vitol et al., 2006). A multiplex PCR assay that differentiates between RDRio and non-RDRio strains (Gibson et al., 2008; Lazzarini et al., 2008) was performed on all isolates from 2005, Cluster VI isolates from 2001–2004, and a representative but randomized selection (10%) of isolates from non-Cluster VI groups from 2001–2004 (LAM strains are known to group in Cluster VI) (Gibson et al., 2008).
Demographic and clinical case data was obtained from the NYC TB Registry (Table 4). Drug use was consolidated as one variable (yes if any injection drug use, non-injection cocaine use, or non-injection drug use), and non-US country of birth was categorized into region of birth: Latin American-Caribbean (included Puerto Rico for purposes of this study only), Asia (East Asia and South East Asia), and Other (including Canada, Europe, and world regions not already stated).
Isolates with identical IS6110-RFLP fingerprints (with identical spoligotyping if ≤ 5 IS6110 bands) were considered clustered (Alland et al., 1994; Maguire et al., 2002). The proportion clustered was calculated, and recent transmission was estimated by the “n-1” method (Small et al., 1994; Murray and Alland, 2002). The secondary case rate [number of secondary cases / (index + unique cases)] (Borgdorff et al., 1998) was calculated for RDRio and other major lineages. The first chronological case cultured was identified as the “index case.”
Categorical variables were compared for different lineages and sublineages using Pearson’s chi-square. Fisher’s exact test was used for small cell counts. The Jonckheere-Terpstra test (Jonckheere, 1954) was used to test for age trend. The proportion of RDRio among TB cases in children (age 0–10) was compared to the proportion of RDRio in older TB cases (age 11–100). All Mycobacterium bovis cases were excluded from the subanalysis, as the M. bovis-infected children were part of a food borne outbreak (CDC, 2005). Confidence intervals for the secondary case rate with a given family were calculated using the bootstrap method with 5000 resamples (Efron and B, 1982), noting that this rate formulation is not based on standard binomial distribution. To characterize cases harboring RDRio compared to non-RDRio M. tuberculosis, we examined the demographic, clinical and social history variables using univariate and multivariate logistic regression models, from which unadjusted and adjusted odds ratios along with the 95% confidence intervals were estimated. Statistical analysis was performed using SAS (version 9.1, SAS Systems, Cary, NC).
The analysis included 3847 unique-case M. tuberculosis strains from 2001 to 2005. There was a widespread distribution of lineages, including LAM (n = 666, 17.3%), W/Beijing (n = 581, 15.1%), Haarlem (n = 626, 16.3%), T (n = 568 14.8%), Central Asian (CAS, (n = 182, 4.7%), and East African-Indian (EAI, n = 304, 8%) (Table 1). We found 7.9% (n = 302) of the NYC M. tuberculosis isolates were RDRio strains. Of these RDRio isolates, 289 (95.7%) had been previously categorized as LAM while 13 (4.3%) had been previously categorized as “non-LAM”. Upon further genetic analysis, the latter 13 isolates each had a LAM-defining Ag85C103 SNP (Gibson et al., 2008); thus confirming that all RDRio isolates are members of the LAM family (Gibson et al., 2008). Furthermore, this study also affirmed that all RDRio strains belong to the SNP Cluster VI grouping (Gutacker et al., 2002).
Fifty-one percent of the RDRio strains were part of a cluster, a proportion higher than the phylogenetically related non-RDRio LAM (42.6%, p=0.035) (Table 2). RDRio strains also had a greater degree of clustering than other prominent lineages, including W/Beijing, but not the Haarlem family (Table 2). Because the index case could represent reactivation, we next used the “n-1” method to estimate recent transmission (Small et al., 1994; Murray and Alland, 2002). By this method, RDRio strains showed a greater proportion of clustering than the non-RDRio LAM and each of the unrelated families, excluding Haarlem (Table 2). A higher secondary case rate (Borgdorff et al., 2000) was also found for the RDRio sublineage than non-RDRio LAM strains and other prominent lineages (except Haarlem) (Table 3). Finally, RDRio caused 19% (7/37; p=0.02) of all TB cases in children 10 years and younger, a proxy marker for recently transmitted TB. A trend was observed in the prevalence of RDRio across age groups using the Jonckheere-Terpstra test, (p=0.03).
Multiple logistic regression analysis found a statistically significant association between TB cases with RDRio and race/ethnicity, year of isolation (increased over study period), region of birth, and drug resistance (Table 4). TB amongst Blacks and Hispanics showed, respectively, nearly 2.3 (95% CI: 1.17–4.63) and 3.3 (95% CI: 1.60–6.99) times greater odds of having an RDRio strain, even after adjusting for birth in Latin American-Caribbean countries along with other factors. INH resistance was associated with RDRio (OR 1.65 [95% CI: 1.09–2.51]), as was cavitary disease (though not statistically significant on multivariate analysis (OR 1.35 [95% CI: 0.98–1.87])).
Further descriptive exploratory analyses of transmission patterns revealed that TB caused by RDRio was associated with case race/ethnicity and birthplace, with differences noted in the percentage of Hispanics and Blacks harboring clustered or non-clustered (unique) RDRio strains. Of the 154 cases with a clustered RDRio strain, 145 (94.1%) were Hispanic or non-Hispanic Black. Of the 93 clustered Hispanic cases, 78 (83.9%) were foreign-born from Latin-American or Caribbean regions. In contrast, 33 (64.7%) of the 51 clustered non-Hispanic Black cases were US-born. Among 148 unique “non-clustered” RDRio case-strains, 67 (85.9%) Hispanics and 24 (43.6%) Blacks were born in Latin American or Caribbean countries. Differences in age at presentation were also noted between US-born Hispanic and Black RDRio TB cases – the median age was 24 years for Hispanics (from n = 25 with the range of 0–52) and 50 years for Blacks (n = 52, range of 7–97).
We also looked at the demographic characteristics of each individual RDRio cluster. Of 43 RDRio clusters, 5 (11.6%) consisted of US-born cases only, 18 (41.9%) consisted of foreign-born cases only and 20 (46.5%) consisted of mixed (both US- and foreign-born) cases. Thirty of the 43 (69.8%) clusters involved only Hispanic (n=21) or Black (n=9) TB cases; while the remaining 13 (30.2%) clusters contained at least one Hispanic and/or Black TB case. Four of the US-born-only clusters were exclusively comprised of Black or Hispanic cases (n=3 or n=1, respectively), with a single cluster involving both Black and Hispanic cases. Three Hispanic-only, 2 Black-only, and 2 multi-race clusters contained a TB case age 10 years or younger (both children in the multi-race clusters were Hispanic). Overall, these data suggest distinct transmission patterns for different clusters of RDRio strains that may be host ethnicity related. Among the foreign-born only clusters, 12 (66.7%) were constituted entirely of cases originating from the same country (Dominican Republic [n clusters=5], Ecuador [n = 3], Mexico [n = 2], Barbados [n = 1], and Puerto Rico [n = 1]). These data add to the number of countries from which RDRio has previously been associated (Gibson et al., 2008). With respect to the mixed origin RDRio clusters, the index case was foreign-born in 15 (75%) and US-born in 5 (25%).
The RDRio deletion is thought to have arisen from a homologous recombination event between non-mobile genes that could have theoretically occurred more than once as homoplastic events in strains from different lineages (Lazzarini et al., 2007). In this study, we confirmed that RDRio is strictly a sublineage of the LAM family and further showed that it is enfolded within the broader SNP Cluster VI grouping. The localization of RDRio strains to a single family suggests that the global distribution of RDRio strains (Gibson et al., 2008) is the result of the dissemination of a single successful clonal sublineage.
A diverse M. tuberculosis clade distribution was observed in this study (n=3847, 2001 to 2005) which is consistent with the M. tuberculosis families found in other cities with significant and diverse immigrant communities (Hirsh et al., 2004; Gagneux et al., 2006; Reed et al., 2009) (Supplemental Table 1 and 2). We report that the LAM spoligotype family was responsible for the highest proportion by lineage (17.3%) of the NYC TB cases from 2001–2005, consistent with its numerical importance amongst M. tuberculosis families in the SpolDB4 international collection database (Brudey et al., 2006). However, less attention has been dedicated to studying this lineage in part because, unlike W/Beijing in the 1990s, it has not been associated with large outbreaks of multi-drug resistant TB or with vulnerable populations (Frieden et al., 1995; Geng et al., 2002). Moreover, the RDRio sublineage of LAM caused ~8% of TB cases in NYC and 45.3% of all LAM-related TB - a substantial proportion for a single M. tuberculosis sublineage competing for hosts against multiple M. tuberculosis genotypes in a multiethnic environment such as NYC.
In this study, we used complimentary methods to estimate RDRio transmission. Prior to this study, few investigations have calculated secondary case rates for TB (Borgdorff et al., 1998) or have reported M. tuberculosis sublineage-specific rates of clustering (Hanekom et al., 2007b; Iwamoto et al., 2009). The data indicate that RDRio strains may be more transmissible as compared to most other major lineages in NYC, including the phylogenetically related non-RDRio LAM. Heretofore, such an analysis has not been possible as there are few recognized sublineages that make up a sufficient proportion of the other major M. tuberculosis families. These data therefore suggest that RDRio will persist in NYC and increasingly become a source of domestic TB transmission amongst and between foreign- and native-born populations in the US. Of note, strains possessing an RDRio-coincident deletion RD174 had an elevated secondary case rate ratio in San Francisco, CA, (note in ref. de Jong et al., 2008), and higher active TB case rate in RD174 strain-exposed persons in The Gambia compared to those exposed to other lineages (de Jong et al., 2008). Since the RD174 deletion is restricted to RDRio strains (Gibson et al., 2008), the prior reports were likely describing RDRio strains, thereby, supporting our observations.
M. tuberculosis is an exquisitely well-adapted human pathogen as a result of a long co-evolutionary history with its human host (Hershberg et al., 2008; Wirth et al., 2008). Recently, Gagneux et al. (Gagneux et al., 2006) proposed that certain M. tuberculosis lineages have differentially adapted to specific human populations, such that transmission within these ethnic groups may be more efficient. Our data indicates that RDRio exhibits human host associations consistent with its phylogeography. The dual host association (Hispanics and Blacks) exhibited by RDRio in NYC is a key difference from previous studies (Gagneux et al., 2006; Reed et al., 2009) and raises the question of whether genetically distinct RDRio stains, with different host associations, may be circulating. The high proportion of racially segregated Black or Hispanic clusters supports this hypothesis and suggests that separate mini-epidemics of RDRio may be occurring in NYC with distinct transmission dynamics. Intriguingly, in a post-hoc analysis of secondary cases, Hispanics with RDRio strains had a higher proportion of cavities on chest radiographs than those with non-RDRio strains (31.5% vs. 21.8%), whereas non-Hispanics had similar cavitation rates for both (23.3% vs. 21.4%). This suggests possible host-pathogen specific disease manifestations, but is a hypothesis requiring further study (Lazzarini et al., 2007; Lazzarini et al., 2008).
An important limitation of the current study is that we are comparing a genomic deletion-defined sublineage to the less precise IS6110-RFLP and spoligotyping methods to define other lineages. Nevertheless, these other lineages provide a background against which to measure the characteristics of the RDRio sublineage. There may well be sublineages within each of these lineages with higher cluster rates and unique demographic characteristics (Hanekom et al., 2007b). Another limitation is that, although self-reported race/ethnicity is a reasonably good predictor of racial heritage (Rosenberg et al., 2002), some Latin American countries have greatly admixed populations that blur racial lines. We also could not completely exclude all potential confounding factors, including unmeasured social-economic factors associated with both RDRio and with delayed access to health care. However, by comparing the transmissibility of the RDRio sublineage to that of the non-RDRio LAM lineage (from which RDRio was derived) we blunt the effect of some potential confounding factors, such as differing cultural norms and a lack of social admixing among ethnic groups. This has not been possible in prior analyses of other lineages with minimally overlapping phylogeographies (Gagneux et al., 2006; Reed et al., 2009), thus strengthening the validity of our observations. Even though this study lacked comparative pathogenicity data in an animal model, the comparison of multiple M. tuberculosis lineages in an ethnically and racially diverse human population with RDRio, and in particular the closely related non-RDRio LAM strains with RDRio strains, overcomes this perceived limitation. Lastly, the association of clustered IS6110-RFLP strains and recent transmission is presumed in molecular epidemiology studies, but may include cases of coincident reactivation while it may miss clusters where the index or secondary case is not in the database.
In conclusion, the present study is one of the few to link data on molecular epidemiology, case demographics, clinical disease characteristics, and strain genotypes to clarify the transmission dynamics of a M. tuberculosis sublineage within a major metropolitan area (Hanekom et al., 2007b; Iwamoto et al., 2009). The RDRio clade is clearly emerging as an important cause of TB, both internationally and in the US, with an increased propensity to cause cavitary disease and with increased transmission efficiency among US-born Blacks and Hispanics and persons of Latin American-Caribbean heritage. Using RDRio as a model, future studies should aim to enhance our knowledge of the contributions the host and the pathogen on TB transmission and will aid the future targeting and evaluation of public health efforts in TB control.
Funding for this study was provided by National Institutes of Health (NIH) R21 AI063147, and R21 AI063147 (JLH.), NIH Fogarty International Center Training Grant (FICTG) D43 TW00018 under the AIDS International Training and Research Program (Warren D. Johnson and JLH), ICOHRTA AIDS/TB 5 U2R TW006883 (Jose.R. Lapa e Silva and JLH) and grant KL2RR024997 (SAW) of the Clinical and Translation Science Center at Weill Cornell Medical College (CTSC funded by Grant (UL1 RR024996). The authors thank Dr. Barun Mathema at the PHRI TB Center for his critical review of the manuscript. The authors also thank Tracy Agerton, Cynthia Driver, and Sonal Munsiff of the Bureau for Tuberculosis Control, New York City Department of Health and Mental Hygiene for their assistance in facilitating the conduct of study without which this project would not have been accomplished.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
This study was presented in part as an abstract at the 2008 46th Annual IDSA/48th Annual ICAAC Joint Meeting (Abstract F1-2015)