JOURNAL OF CLINICAL MICROBIOLOGY, Oct. 2010, p. 3593–3599 Vol. 48, No. 10 0095-1137/10/$12.00 doi:10.1128/JCM.00430-10 Copyright ? 2010, American Society for Microbiology. All Rights Reserved.
Strains of Mycobacterium tuberculosis from Western Maharashtra,
India, Exhibit a High Degree of Diversity and Strain-Speci，c
Associations with Drug Resistance, Cavitary Disease, and
Treatment Failure †
111342Anirvan Chatterjee,Desiree D’Souza,Tina Vira,Arun Bamne,Gurish T. Ambe,Mark P. Nicol, 1Robert J. Wilkinson,5,6,7 and Nerges Mistry*
1Foundation for Medical Research, 84-A, R. G. Thadani Marg, Worli, Mumbai 400018, India; Division of Medical Microbiology, 2Medical School, University of Cape Town, Observatory, Cape Town 7925, South Africa; Mumbai District TB Control Society 111, 3New Municipal Building, Dr. E. Moses Road, Worli, Mumbai 400018, India; Joint Executive Health Of;cer, F(S) Ward Building, 4Brihanmumbai Municipal Corporation (BMC), Mumbai 400012, India; Division of Medicine, Imperial College London, 5London W2 1PG, United Kingdom; Institute of Infectious Diseases and Molecular Medicine and Department of 6Medicine, Faculty of Health Sciences, University of Cape Town, Cape Town, South Africa; and 7MRC National Institute for Medical Research, Mill Hill, London NW7 1AA, United Kingdom
Received 2 March 2010/Returned for modi，cation 21 June 2010/Accepted 10 August 2010
We performed spoligotyping of Mycobacterium tuberculosis isolates from 833 systematically sampled pulmo- nary tuberculosis (TB) patients in urban Mumbai, India (723 patients), and adjacent rural areas in western India (110 patients). The urban cohort consisted of two groups of patients, new cases (646 patients) and ；rst-time treatment failures (77 patients), while only new cases were recruited in the rural areas. The isolates from urban new cases showed 71% clustering, with 168 Manu1, 62 CAS, 22 Beijing, and 30 EAI-5 isolates. The isolates from ；rst-time treatment failures were 69% clustered, with 14 Manu1, 8 CAS, 8 Beijing, and 6 EAI-5 isolates. The proportion of Beijing strains was higher in this group than in urban new cases (odds ratio [OR], 3.29; 95% con；dence limit [95% CL], 1.29 to 8.14; P 0.003). The isolates from rural new cases showed 69% clustering, with 38 Manu1, 7 CAS, and 1 EAI-5 isolate. Beijing was absent in the rural cohort. Manu1 was found to be more common in the rural cohort (OR, 0.67; 95% CL, 0.42 to 1.05; P 0.06). In total, 71% of isolates were clustered into 58 spoligotypes with 4 predominant strains, Manu1 (26%), CAS (9%), EAI-5 (4%), and Beijing (4%), along with 246 unique spoligotypes. In the isolates from urban new cases, we found Beijing to be associated with multidrug resistance (MDR) (OR, 3.40; 95% CL, 1.20 to 9.62; P 0.02). CAS was found to be associated with pansensitivity (OR, 1.83; 95% CL, 1.03 to 3.24; P 0.03) and cavities as seen on chest radiographs (OR, 2.72; 95% CL, 1.34 to 5.53; P 0.006). We recorded 239 new spoligotypes yet unreported in the global databases, suggesting that the local TB strains exhibit a high degree of diversity.
The resurgence of tuberculosis (TB) fuelled by multidrug simultaneously for long durations, resulting in a set of varied resistance (MDR) and extensive drug resistance has caused strain types (6, 20, 21). signi，cant concern among health care practitioners (36, 37). Studies from India show differential strain predominance There have been renewed efforts to understand the biology of between the southern and northern regions of the country. the pathogen alongside its epidemiology. Such data have come While the central Asian strain (CAS) is dominant in the north, mostly from regions of sporadic incidence or from populations East African Indian strains (EAI) are observed more fre- where the disease is driven by high HIV prevalence (8). Data quently in the southern regions (29). Most studies from India from some of the highest-disease-burden regions, where tuber- as well as speci，cally from Mumbai, India, showed CAS and culosis has remained endemic, are scarce. Manu1 as the predominant spoligotypes along with EAI as a India is one such region where Mycobacterium tuberculosis third large strain lineage (3, 16, 18, 22, 28, 29, 30). Another has remained in equilibrium with the population, resulting in study from a tertiary care center in Mumbai reported a high an area of tuberculosis endemicity (6, 27). Under such condi- proportion of Beijing strains (23%) in a cohort associated with tions, the strain diversity is expected to be different compared a high proportion of MDR (1). Interestingly, TbD1-positive to that for epidemic or sporadic incidents, where a few speci，c strains of tuberculosis (such as EAI) predominate in India,
strain types dominate (8, 23); in a setting where tuberculosis is whereas TbD1-negative strains of M. tuberculosis are more
endemic, the pathogen and the host are expected to evolve common in the rest of the world (11). Although previous studies provided preliminary data from various sites in India, they did not re！ect strain variability from * Corresponding author. Mailing address: Foundation for Medical a single cosmopolitan region. Additionally, studies from Mum- Research, 84-A, R. G. Thadani Marg, Worli, Mumbai 400018, India. bai were biased toward MDR cases (1), had small sample sizes Phone: 91 22 4934989. Fax: 91 22 4932876. E-mail: firstname.lastname@example.org. (16), or were derived from a cohort of chronic (re-treated) TB † Supplemental material for this article may be found at http://jcm cases accessing tertiary care hospitals (22). .asm.org/. Epidemiological studies of M. tuberculosis have been facili- Published ahead of print on 18 August 2010.
3594 CHATTERJEE ET AL. J. CLIN. MICROBIOL.
tated by a variety of genotyping tools. IS6110 restriction frag- ment length polymorphism (RFLP) remains the gold standard due to its high level of discrimination (15) but is time-consum- ing and less suitable in populations with low copy numbers (7). Mycobacterial interspersed repetitive-unit–variable-number tandem-repeat (MIRU-VNTR) typing (10, 19, 31) is a high- throughput and discriminatory method, but the best combina- tion of MIRU loci is yet to be achieved, and combinations may differ between populations (33). We used spoligotyping as a primary ，ngerprint method, due to its relatively high through- put nature. Spoligotyping has a lower discriminatory power than MIRU-VNTR typing, making it less suitable for deter- mining strain transmission. However, spoligotyping served as a useful primary ，ngerprinting tool allowing comparisons of strain types from strains around the world through updated global databases (5, 17, 35). In this study, we describe the distribution of strain genotypes from a systematic collection of strains from urban Mumbai and two neighboring rural areas. Mumbai is a location where a con！uence of people from all parts of the country live in poor, congested neighborhoods with high population densities, up to 64,168 people per square kilometer in one of the city wards as
per the 2001 census (26). These conditions, coupled with a high
proportion of MDR cases in the region (2, 9), were cause for
concern and underlined the need for more information on FIG. 1. Geographical demarcations of India. (The map was created local circulating strains. We wished to determine the distribu- using MapXL Maps of India, v.6 [Compare Infobase Limited].) tion of strains among newly diagnosed TB patients in the region to extend previous observations (16, 22). This study describes the spoligotypes present in the cohort All rural patients were new cases sampled at onset of therapy. Patients were and the extent of their clustering. We further analyzed the recruited from April 2004 to September 2007 at RNTCP DOTS centers. association of spoligotypes with other parameters, namely, age, Inclusion criteria for patients were (i) smear positivity, (ii) age from 15 to 70 gender, geographical origin of the host, radiology, and multi- years, (iii) residency in Mumbai for at least 3 years immediately prior to diag- drug resistance, to obtain deeper insights into strain behavior. nosis, and (iv) residence in the same area as the health posts where treatment was sought. Sputum-negative cases were excluded, as we were unable to perform sputum MATERIALS AND METHODS culture on all patients, due to study constraints. Patients who had not resided in Location of study. The Revised National Tuberculosis Control Programme the area for at least 3 years were excluded, as we were interested primarily in (RNTCP) in Mumbai is implemented in individual wards of the city. Since this strains causing disease in the resident population of Mumbai. Since the study study was part of a larger epidemiological project assessing transmission of MDR aimed to identify strains circulating in the region during the study period, pre- TB in a setting of TB endemicity, 4 centrally located wards (F/N, G/N, H/E, and viously treated patients were excluded to eliminate relapse cases, which are less K/E) characterized by a high sputum-positive case load, with moderately subop- likely to relate to a current transmission event. Patients with a history of TB or timal cure rates ranging between 78 and 81%, were selected (RNTCP quarterly antituberculosis therapy were determined through interview and scrutiny of reports 2001; can be sourced from email@example.com). As far as could be district TB registers and patients defaulting during therapy (in the case of ascertained, there was no apparent deviation in the RNTCP functioning in these ，rst-time treatment failures). As far as could be ascertained, the ，rst-time treat- wards compared to that in the other wards in Mumbai. A signi，cant proportion ment failures had not received any antituberculosis therapy prior to the current of the resident population of these 4 selected wards belonged to the middle or episode. Patients were recruited after informed consent and referred for HIV lower socioeconomic class and resided in informal housing in slums. Cumula- counseling and testing. tively, the 4 wards covered 38 DOTS (directly observed therapy short course) Clearance for this study was obtained from the Foundation for Medical Re- centers, with a population of 3 million. search (FMR) institutional ethics committee (20.07.2001/01). In addition, we sampled patients from rural TB units (located 250 and 280 km Patient investigations. Patient demographic details, including age, gender, and from Mumbai), covering a population of 860,000. This facilitated studying spo- geographical origin, were recorded. Since many patients were found to be mi- ligotypes from a rural cohort which, as opposed to Mumbai, had a homogeneous grants from other parts of the country, they were analyzed based on 3 groups: population comprising an indigenous community with little or no migration from south India, north India, and Maharashtra (considered the native population) other parts of India. Unlike the rural cohort, that from the Mumbai region (Fig. 1). included a large migrant population from other parts of India (37%) (26). We All patients recruited were subjected to radiological examination and were could thus compare strain diversities in homogenous (rural) and cosmopolitan binomially classi，ed into groups by presence or absence of cavities. (Mumbai, urban) cohorts in the same geographical location. Additionally, there Sample processing and culture. Sputum samples (one per patient) were col- exists a movement of people from these rural areas to Mumbai, in the form of lected in cetylpyridinium chloride (CPC)-NaCl vials as previously described (9). unskilled industrial labor, which provided the opportunity to study the overlap of These samples were then concentrated by Petroff’s method (4). The concen- TB strains between the two locations. trated samples were inoculated onto solid Lowenstein-Jensen slants (Hi-Media, Case de；nition. Sputum-positive new cases of TB were broadly classi，ed into India) and incubated at 37?C until growth was observed. two groups: (i) new cases sampled at onset of CAT1 therapy, a regimen of DNA extraction. A loopful of culture was used to extract DNA by a standard 2(isoniazid [H]-ethambutol [E]-rifampin [R]-pyrazinamide [Z])3 4(H-R)3, cetyltrimethylammonium bromide (CTAB) mycobacterial DNA extraction pro- comprising 2 months of H-E-R-Z thrice weekly followed by 4 months of H-R cedure, followed by phenol chloroform puri，cation (13, 14). thrice weekly), and (ii) ，rst-time treatment failures, or new cases who remained Spoligotyping. Spoligotyping was conducted as previously described (12). sputum smear positive at the ，fth month after commencement of CAT1 therapy. Clusters were de，ned as at least 2 isolates with identical spoligotypes.
VOL. 48, 2010 SPOLIGOTYPING M. TUBERCULOSIS STRAINS IN WESTERN INDIA 3595
TABLE 1. Patient distribution in the cohort across the different locations and patient types
Spoligotype clustering No. of isolates of major spoligotype
Patient study group Ward or TB No. (%) of No. of unique unit and cohort Total Manu1 CAS Beijing EAI-5 clustered isolates isolates Treatment failures 10 F/N 20 (67) 30 5 2 3 3 Mumbai (urban) G/N 4 (50) 4 8 0 2 1 0 H/E 13 (65) 7 20 5 2 3 1 K/E 16 (84) 3 19 4 2 1 2 Total 53 (69) 24 77 14 8 8 6 New cases 194 (72) 74 268 64 28 14 14 F/N Mumbai (urban) G/N 78 (70) 34 112 37 8 1 4 H/E 86 (74) 31 117 32 9 1 3 K/E 100 (67) 49 149 35 17 6 9 Total 458 (71) 188 646 168 62 22 30 Narayangaon 46 (70) 20 66 24 3 0 1 Rural Bhor 30 (68) 14 44 14 4 0 0 Total 76 (69) 34 110 38 7 0 1
587 (71) 246 833 220 77 30 37 Total
A random selection of samples (DNA extracted from 80 isolates in Tris-EDTA more than 2 weeks, and 14 (6%) who resided outside the study [TE] in freezer packs) was sent to the Institute of Infectious Diseases and area. Twenty-seven patients (11%) refused to participate in the Molecular Medicine, University of Cape Town (IIDMM, UCT), for single- study. blinded quality control. (ii) Rural. The rural cohort consisted of 110 patients out of Drug susceptibility testing. Drug susceptibility testing for the four ，rst-line drugs rifampin (R), isoniazid (H), ethambutol (E), and pyrazinamide (Z) was 262 screened (42%).Thus, a total of 833 isolates from urban performed by the radiorespirometric Buddemeyer technique (a manual modi，- (n 723) and rural (n 110) patients were spoligotyped cation of the BACTEC 460 technique) (9). Strains resistant to at least H and R (Tables 1, 2, 3, and 4). were classi，ed as multidrug resistant (MDR). Resistance to H, R, and an addi- Spoligotyping. The four major clusters, designated C1, C2, tional ，rst-line drug was represented as MDR . Ten percent of the isolates were C15, and C24, contained 77, 30, 220, and 37 isolates, respec- sent to the Swedish Institute for Infectious Disease Control (supranational ref- erence laboratory), Stockholm, for external quality assurance by the BACTEC tively (Fig. 2). The clusters were labeled C1 through C58 se- method. Kappa scores showed good agreement for H (0.76) and R (0.77) (9). quentially as they were formed during the study period. Statistical analysis. Multivariate analysis (binary logistic regression) was per- When all of the spoligotypes were compared to the interna- formed with the spoligotype as the dependent variable and MDR, susceptibility, tional spoligotype database SpoldB4 (5), the shared types (ST) HIV status, and cavitation in the new cases (，rst-time treatment failures ex- cluded) as covariates. A chi-square test was performed to ，nd associations and labels were C1, CAS1_Delhi (ST26); C2, Beijing (ST1); between spoligotype and (i) patient subgroup (，rst-time treatment failures, ur- C15, Manu1 (ST100); and C24, EAI-5 (ST236). Therefore, ban new cases, and rural cases) or (ii) region of origin. Manu1 emerged as the largest strain type, infecting 26.4% of All patient data were entered and maintained using SPSS v10.0 (SPSS, the urban and rural cohorts (see Table S1 in the supplemental Inc.). Analyses were done with SPSS and Microsoft Of，ce Excel 1997 (Mi- crosoft Corporation). Spoligotype comparison to SpoldB3 (35) (http://cgi2.cs material). .rpi.edu/ bennek/SPOTCLUST.html) and SpoldB4 (17) (http://www.pasteur New cases (urban). The urban cohort consisted of 646 -guadeloupe.fr:8081/SITVITDemo/) was done using tools available on the isolates from new cases sampled at onset of therapy. Seventy- respective websites. one percent (458/646) of these were found to be clustered into 56 spoligotypes (Table 3). Manu1 was the largest cluster, with RESULTS 168 (26%) isolates (Table 1). We found 62 (9.5%) CAS iso-
lates, 22 (3.4%) Beijing isolates, and 30 (4.6%) EAI-5 isolates Patient selection. (i) Urban. We screened 1,136/2,184 (52%) in the cohort (Table 1). Although this was the largest subgroup new cases who presented to the RNTCP for diagnosis between April 2004 and September 2007. We included 646 new cases in the urban cohort from the 1,136 screened, based on our inclu- sion criteria. The 490 patients excluded from the study con- TABLE 2. Distribution of clustered strains in the total cohort sisted of 135 (28%) with prior antituberculosis treatment, 222
(45%) who had taken more than 5 doses of antituberculosis No. of clusters No. of strains with this No. of isolates with this characteristic/total no. therapy before sampling, and 24 (5%) who resided outside the within cluster characteristic of isolates (%) study area. One hundred nine patients (22%) refused to par- 8 421/833 (51) ticipate in the study. During the same period, we screened 318 10 10 50 166/833 (20) treatment failures, of which 77 were included. The 241 ex- 5 41 108/833 (13) cluded patients consisted of 86 (36%) with prior antitubercu- 2 23 46/833 (0.06) losis therapy, 114 (47%) with an interruption in treatment of
3596 CHATTERJEE ET AL. J. CLIN. MICROBIOL.
TABLE 3. All spoligotypes in different patient types TABLE 4. Distribution of major spoligotypes based on
geographical origin of patient No. of isolates in patient study group aCluster or unique Result for patients with GOof: Total spoligotype Urban treatment Urban new Rural new Spoligotype category and Total failures cases cases Eastern Western Southern Maharashtra Northern parameter India India India C1 8 62 7 77 India C2 8 22 30 Manu1 2 C3 1 2 5 59 130 5 11 15 220 No. of isolates C4 8 2 10 59 % within cluster 27 2 5 7 100 C5 8 8 % within GO 28 25 33 28 21 26 1 1 C6 2 % of total 16 7 1 1 2 26 9 C7 1 11 1 C8 5 5 CAS 2 C9 2 32 31 0 7 7 77 No. of isolates 2 3 C10 5 % within cluster 42 40 0 9 9 100 1 C11 18 17 % within GO 7 13 0 18 10 9 2 C12 1 15 18 % of total 4 0 1 1 9 4 C13 2 2 C14 2 2 EAI-5 14 38 C15 168 220 25 0 1 3 37 8 No. of isolates C16 3 2 1 % within cluster 68 22 0 3 8 100 1 C17 2 2 5 % within GO 5 0 3 4 4 3 C18 1 2 1 % of total 3 1 0 0 0 4 C19 2 2 C20 3 3 Beijing C21 2 2 12 13 1 1 3 30 C22 3 4 No. of isolates 1 % within cluster 40 43 3 3 10 100 C23 3 3 % within GO 3 6 7 3 4 4 C24 6 30 1 37 % of total 1 2 0 0 0 4 C25 2 2 C26 2 2 C27 3 3 Other clusters C28 2 2 66 4 9 27 223 No. of isolates 117 C29 2 2 % within cluster 53 30 2 4 12 100 C30 3 3 % within GO 25 28 27 23 37 27 C31 1 2 3 % of total 14 8 1 1 3 27 C32 1 3 4 C33 2 2 Unique C34 3 4 1 spoligotype C35 2 2 153 60 5 10 No. of isolates 18 246 C36 3 3 62 24 2 4 7 100 % within cluster 3 C37 4 7 33 25 33 26 25 30 % within GO C38 4 4 18 7 1 1 2 30 % of total C39 4 4 1 C40 2 3 Total C41 4 4 469 237 15 39 73 833 No. of isolates C42 3 3 % within cluster 56 29 2 5 9 100 C43 3 3 100 100 100 100 100 % within GO 100 C44 2 2 % of total 56 29 2 5 9 100 C45 4 4 C46 2 2 C47 a GO, geographical origin. 4 4 8 C48 2 2 1 1 C49 5 7 C50 1 2 1 C51 1 1 2 1.29 to 8.14; P 0.003) (Table 5). This subgroup consisted of 1 C52 3 4 3 C53 1 4 8 both the isolates that formed the C9 cluster not found in new C54 2 2 cases and 1 isolate of the C6 cluster (Table 3). 1 C55 1 2 C56 1 2 1 New cases (rural). The 110 isolates in the rural cohort were C57 2 2 divided into 21 clusters and 34 unique spoligotypes. The 21 C58 2 2 24 34 Unique spoligotype 188 246 clusters comprised 69% (76/110) of the rural isolates and were
also seen in the urban cohort. Manu1 was the largest cluster, Total 646 833 77 110 with 38 (34.5%) of the isolates, followed by 7 CAS isolates
(6.3%) and 1 EAI-5 isolate. Additionally, we found no Beijing
strains among the 110 isolates (Table 1).
of patients, it did not contain 2 spoligotypes (C9 and C6) In total, we found 833 spoligotypes, of which 246 were
unique, while the remaining were distributed across 58 spoli- (Table 3).
gotype clusters. A total of 70.5% of strains were clustered. The First-time treatment failures. Seventy-seven treatment fail-
proportions of clustered strains were similar in the urban and ures sampled at the end of 5 months of treatment yielded 24
rural cohorts (70.6% and 69.1%, respectively). unique spoligotypes and a clustering of 69% (Tables 1 and 3).
Although 71% of the isolates were present in clusters, a The 53 clustered strains were divided into 18 spoligotypes.
large number of these clusters had few isolates in them. Of the Manu1 was the largest cluster, with 14 isolates (18%), followed
by CAS and Beijing, with 8 isolates each (10%). EAI-5 was the 587 clustered strains, 421 (72%) were in clusters with greater fourth largest cluster, with 6 isolates (8%) (Table 1). The than 10 isolates in them. Fifty-one percent of the isolates proportion of Beijing isolates in the treatment failures was formed only 8 clusters. Twenty-three clusters had only 2 iso- found to be signi，cantly higher than that in the urban new lates in them (Table 2).
cases (odds ratio [OR], 3.29; 95% con，dence limit [95% CL], Additionally, we have reported 239/304 (79%) new spoligo-
VOL. 48, 2010 SPOLIGOTYPING M. TUBERCULOSIS STRAINS IN WESTERN INDIA 3597
FIG. 2. Spoligotype proportions.
types (22 clustered and 217 unique) which may be added to for the rest of the cohort (OR, 2.72; 95% CL, 1.34 to 5.53; P
existing databases to provide a better representation of strains 0.006) (Table 5).
from a high-burden region (see Table S1 in the supplemental (iii) Geographical origin of patient. On analyzing the spoli- material). gotypes for association with speci，c host background (Table A concordance of 97% was found between spoligotyping 4), we found a signi，cantly high number of CAS strains in results from our laboratory and IIDMM, UCT. isolates from people originating from the northern regions of
India compared to the numbers in the rest of the cohort (OR, Spoligotype associations (new cases). (i) Multidrug resis-
1.84; 95% CL, 1.08 to 2.99; P 0.02) (Table 5) and in the tance. We did not observe any signi，cant difference in the
clustering percentages between treatment failures and new indigenous population (OR, 2.05; 95% CL, 1.18 to 3.56; P
cases (68.8% and 70.6%, respectively) (data not shown) or 0.006). We also found a higher proportion of Manu1 strains in across different drug susceptibility pro，les (sensitive, 68%; the rural population than in the urban new cases (OR, 0.67; MDR, 68%; single-drug resistant, 75%; other resistance, 73%) 95% CL, 0.42 to 1.05; P 0.06) (data not shown) and a (data not shown). signi，cantly higher proportion than in the urban cohort (OR,
1.57; 95% CL, 0.41 to 1.00; P 0.04). Analyses of associations between spoligotype and drug re-
sistance showed a signi，cantly higher proportion of Beijing No signi，cant associations between spoligotype and age or
gender were seen. strains than other strain types among the MDR isolates (OR,
3.40; 95% CL, 1.20 to 9.62; P 0.02). We also found a signif- (iv) HIV status. Of the 833 patients, HIV status was avail-
able for 722 individuals. Of these, 32 (4.5%) had tested posi- icantly higher proportion of CAS strains than other strain types
in the pansensitive group (OR, 1.83; 95% CL, 1.03 to 3.24; P tive. No association between HIV status and spoligotype was
found. 0.03) (Table 5). Beijing strains were also found to be signi，-
cantly associated with treatment failures (OR, 3.29; 95% CL,
1.29 to 8.14; P 0.003), which was independent of MDR, as DISCUSSION found by multivariate analysis. This to the best of our knowledge is the largest community- (ii) Extent of radiographic disease. We further observed
based molecular ，ngerprinting study of M. tuberculosis with a that infection with CAS was signi，cantly associated with the
well-characterized cohort of new pulmonary tuberculosis cases presence of cavities on chest radiographs, compared to results
TABLE 5. Signi，cant associations between spoligotype and biological parameter c No. of isolates of spoligotype OR 95% CL P value Parameter Manu1 CAS Beijing EAI-5
aRural cohort 38* 7 0 1 1.57 0.41–1.00 0.04 bTreatment failures 14 8 8* 6 3.29 1.29–8.14 0.003 b% of isolates MDR within a cluster (urban new cases) 20 17 41* 27 3.40 1.20–9.62 0.02 b% of isolates pansensitive within a cluster (urban new cases) 33 53* 41 47 1.83 1.03–3.24 0.03 a% of patients with lung cavities within a cluster 54 83* 62 56 2.72 1.34–5.53 0.006 b% of patients from north India within a cluster 27 40* 43 22 1.8 1.08–2.99 0.02
a Chi-square test. b Binary logistic regression. c *, proportion is signi，cantly high compared to those of other spoligotypes.
3598 CHATTERJEE ET AL. J. CLIN. MICROBIOL.
from India (1, 2, 3, 11, 16, 22, 28, 29, 30). We included only elsewhere in India (CAS and EAI-5) in the cosmopolitan ur- sputum-positive patients to enhance the possibility of typing ban cohort.
The preliminary inferences from this study plead for a more actively transmitting strains. Although the exclusion of spu-
extensive analysis of the data to study the variability of M. tum-negative cases could potentially have introduced a bias in
tuberculosis strains and their transmission dynamics. the strain composition, we were unable to perform sputum
The study represents a ，rst step toward amalgamating clin- culture on all patients due to study constraints. We demon-
ical, genetic, and social factors that are intrinsic to the evolu- strated a predominance of Manu1 along with a high proportion tion of the pathogen. of CAS strains in our cohort, versus the predominance of Harlem and LAM strains in Africa and South America, Beijing ACKNOWLEDGMENTS and T strains in the Americas, the Beijing strain in East Asia,
and the T strain in Europe (5, 32). Concurrent with earlier This project was supported by The Wellcome Trust (CRIG grant reports from India (28) and speci，cally Mumbai (16), we found 073138/Z/03/A). We thank our clinical consultant, Yatin Dholakia, our statistical a low proportion of the Beijing strain in our cohort. This is consultant, K. Ramchandran, and our ，eld operation consultant, different from the 23% proportion of Beijing reported by a Sheela Rangan, for their expert advice. We thank our ，eld workers tertiary care center in Mumbai catering to a more af！uent who collected the samples and also interviewed all of the patients. We population (1). The high incidence of Beijing in that study may thank the late N. H. Antia, the founder and director of the Foundation for Medical Research, who paved the way for doing meaningful re- be attributed to the selection bias toward MDR TB patients search combining science and public health. We also extend our sin- and also to a signi，cantly higher proportion of migrant Tibetan cerest thanks to all of the patients, without whose support this project populations accessing the tertiary care center (personal com- would never have been possible. munication from Camilla Rodriguez, Hinduja Hospital, Mum- REFERENCES bai). Additionally, we found a high strain diversity, with a large 1. Almeida, D., C. Rodrigues, T. F. Ashavaid, A. Lalvani, Z. F. Udwadia, and A. number of small clusters, as well as a signi，cant proportion Mehta. 2005. High incidence of the Beijing genotype among multidrug- (79%) of strains hitherto unreported in the global databases. resistant isolates of Mycobacterium tuberculosis in a tertiary care center in In comparing the spoligotype proportions reported from Mumbai, India. Clin. Infect. Dis. 40:881–886. 2. Almeida, D., C. Rodrigues, Z. F. Udwadia, A. Lalvani, G. D. Gothi, P. Mehta, other studies in India (3, 28, 29) to our observations, we found and A. Mehta. 2003. Incidence of multidrug-resistant tuberculosis in urban that our cohort had a higher proportion of Manu1 (26% versus and rural India and implications for prevention. Clin. Infect. Dis. 36:e152– e154. 7%), a lower proportion of EAI-5 (4% versus 20 to 45%), and 3. Arora, J., U. B. Singh, N. Suresh, T. Rana, C. Porwal, A. Kaushik, and J. N. a comparable proportion of CAS (9% versus 5 to 43%). These Pande. 2009. Characterization of predominant Mycobacterium tuberculosis results may indicate that different strains predominate in dif- strains from different subpopulations of India. Infect. Genet. Evol. 9:832– 839. ferent regions of the country. 4. Baker, J. F., and R. E. Silverton. 1978. Introduction to medical laboratory The low proportion of Beijing in the urban cohort and its technology, 5th ed. Butterworths, London, United Kingdom. 5. Brudey, K., J. R. Driscoll, L. Rigouts, W. M. Prodinger, A. Gori, S. A. absence in the rural cohort suggest that this strain remains Al-Hajoj, C. Allix, L. Aristimuno, J. Arora, V. Baumanis, L. Binder, P. uncommon among the general population in this part of India. Cafrune, A. Cataldi, S. Cheong, R. Diel, C. Ellermeier, J. T. Evans, M. The independent association of Beijing with MDR and Fauville-Dufaux, S. Ferdinand, D. Garcia de Viedma, C. Garzelli, L. Gaz- zola, H. M. Gomes, M. C. Guttierez, P. M. Hawkey, P. D. van Helden, G. V. treatment failure cases (as brought out by multivariate analy- Kadival, B. N. Kreiswirth, K. Kremer, M. Kubin, S. P. Kulkarni, B. Liens, sis) supports earlier ，ndings (23, 25). In contrast, CAS was T. Lillebaek, M. L. Ho, C. Martin, I. Mokrousov, O. Narvskaia, Y. F. Ngeow, L. Naumann, S. Niemann, I. Parwati, Z. Rahim, V. Rasolofo-Razanam- found to be associated with drug sensitivity and with cavitation. parany, T. Rasolonavalona, M. L. Rossetti, S. Rusch-Gerdes, A. Sajduda, S. These observations suggest that modern strains like Beijing Samper, I. G. Shemyakin, U. B. Singh, A. Somoskovi, R. A. Skuce, D. van may have evolved more under drug pressures, leading to an Soolingen, E. M. Streicher, P. N. Suffys, E. Tortoli, T. Tracevska, V. Vincent, T. C. Victor, R. M. Warren, S. F. Yap, K. Zaman, F. Portaels, N. Rastogi, and accumulation of drug-resistant mutations, while ancestral C. Sola. 2006. Mycobacterium tuberculosis complex genetic diversity: mining strains, like CAS, may have evolved to cause cavitation and the fourth international spoligotyping database (SpolDB4) for classi，cation, population genetics and epidemiology. BMC Microbiol. 6:23. hence increase the opportunity for transmission (24, 34). 6. Coleman, P. G., B. D. Perry, and M. E. Woolhouse. 2001. Endemic stabil- Conclusion. Results from this study raise issues that may be ity—a veterinary idea applied to human public health. Lancet 357:1284– typical for a cosmopolitan population in which TB is endemic. 1286. 7. Cronin, W. A., J. E. Golub, L. S. Magder, N. G. Baruch, M. J. Lathan, L. N. The constant exposure of the host to the pathogen (as in a Mukasa, N. Hooper, J. H. Razeq, D. Mulcahy, W. H. Benjamin, and W. R. setting of disease endemicity) has probably resulted in the high Bishai. 2001. Epidemiologic usefulness of spoligotyping for secondary typing strain diversity. In contrast to the variability, the predominance of Mycobacterium tuberculosis isolates with low copy numbers of IS6110. J. Clin. Microbiol. 39:3709–3711. of Manu1 (highest proportion ever reported) is indicative of a 8. Day, C., and A. Gray. 2006. Health and related indicators, p. 369–506. In P. persistence of local strains. Interestingly, the high proportion Ijumba and A. Padarath (ed.), South African health review 2006. Health Systems Trust, Durban, South Africa. of Manu1 in the urban cohort was signi，cantly lower than that 9. D’Souza, D. T., N. F. Mistry, T. S. Vira, Y. Dholakia, S. Hoffner, G. Pasvol, in the more homogenous rural cohort. Concurrently, Beijing M. Nicol, and R. J. Wilkinson. 2009. High levels of multidrug resistant (considered to be an imported strain), displaying a low pro- tuberculosis in new and treatment-failure patients from the Revised National Tuberculosis Control Programme in an urban metropolis (Mumbai) in West- portion in the urban cohort, was absent in the more homoge- ern India. BMC Public Health 9:211. neous rural cohort. 10. Frothingham, R., and W. A. Meeker-O’Connell. 1998. Genetic diversity in the Mycobacterium tuberculosis complex based on variable numbers of tan- The increasing predominance of one strain (Manu1) along dem DNA repeats. Microbiology 144:1189–1196. with the decreasing penetrance of another (Beijing) as the 11. Gutierrez, M. C., N. Ahmed, E. Willery, S. Narayanan, S. E. Hasnain, D. S. homogeneity of the population increased from urban to rural Chauhan, V. M. Katoch, V. Vincent, C. Locht, and P. Supply. 2006. Pre- dominance of ancestral lineages of Mycobacterium tuberculosis in India. probably re！ects the effect of migration. This is further sub- Emerg. Infect. Dis. 12:1367–1374. stantiated by the signi，cant presence of strains predominant 12. Kamerbeek, J., L. Schouls, A. Kolk, M. van Agterveld, D. van Soolingen, S.
VOL. 48, 2010 SPOLIGOTYPING M. TUBERCULOSIS STRAINS IN WESTERN INDIA 3599
Kuijper, A. Bunschoten, H. Molhuizen, R. Shaw, M. Goyal, and J. van associates with immune subversion. Proc. Natl. Acad. Sci. U. S. A. 103: Embden. 1997. Simultaneous detection and strain differentiation of Myco- 15594–15598. bacterium tuberculosis for diagnosis and epidemiology. J. Clin. Microbiol. 25. Parwati, I., B. Alisjahbana, L. Apriani, R. D. Soetikno, T. H. Ottenhoff, A. G. 35:907–914. van der Zanden, J. van der Meer, D. van Soolingen, and R. van Crevel. 2010. 13. Kolk, A. H., A. R. Schuitema, S. Kuijper, J. van Leeuwen, P. W. Hermans, Mycobacterium tuberculosis Beijing genotype is an independent risk factor for J. D. van Embden, and R. A. Hartskeerl. 1992. Detection of Mycobacterium tuberculosis treatment failure in Indonesia. J. Infect. Dis. 201:553–557. tuberculosis in clinical samples by using polymerase chain reaction and a 26. Registrar General and Census Commissioner of India. 2001. Census of nonradioactive detection system. J. Clin. Microbiol. 30:2567–2575. India. Registrar General and Census Commissioner of India, New Delhi, 14. Kox, L. F., D. Rhienthong, A. M. Miranda, N. Udomsantisuk, K. Ellis, J. van India. http://www.censusindia.net/. Leeuwen, S. van Heusden, S. Kuijper, and A. H. Kolk. 1994. A more reliable 27. Rothschild, B. M., and R. Laub. 2006. Hyperdisease in the late Pleistocene: PCR for detection of Mycobacterium tuberculosis in clinical samples. J. Clin. validation of an early 20th century hypothesis. Naturwissenschaften 93:557– Microbiol. 32:672–678. 564. 15. Kremer, K., D. van Soolingen, R. Frothingham, W. H. Haas, P. W. Hermans, 28. Sharma, P., D. S. Chauhan, P. Upadhyay, J. Faujdar, M. Lavania, S. Sa- C. Martin, P. Palittapongarnpim, B. B. Plikaytis, L. W. Riley, M. A. Yakrus, chan, K. Katoch, and V. M. Katoch. 2008. Molecular typing of Mycobacte- J. M. Musser, and J. D. van Embden. 1999. Comparison of methods based rium tuberculosis isolates from a rural area of Kanpur by spoligotyping and on different molecular epidemiological markers for typing of Mycobacterium mycobacterial interspersed repetitive units (MIRUs) typing. Infect. Genet. tuberculosis complex strains: interlaboratory study of discriminatory power Evol. 8:621–626. and reproducibility. J. Clin. Microbiol. 37:2607–2618. 29. Singh, U. B., J. Arora, N. Suresh, H. Pant, T. Rana, C. Sola, N. Rastogi, and 16. Kulkarni, S., C. Sola, I. Filliol, N. Rastogi, and G. Kadival. 2005. Spoligo- J. N. Pande. 2007. Genetic biodiversity of Mycobacterium tuberculosis isolates typing of Mycobacterium tuberculosis isolates from patients with pulmonary from patients with pulmonary tuberculosis in India. Infect. Genet. Evol. tuberculosis in Mumbai, India. Res. Microbiol. 156:588–596. 7:441–448. 17. Liens, B., C. Sola, K. Brudey, N. Rastogi, et al. 2005. A web-site for a global 30. Singh, U. B., N. Suresh, N. V. Bhanu, J. Arora, H. Pant, S. Sinha, R. C. database of Mycobacterium tuberculosis complex spoligotypes and MIRU- Aggarwal, S. Singh, J. N. Pande, C. Sola, N. Rastogi, and P. Seth. 2004. VNTRs (SITVIT), abstr. O1. 26th Annu. Congr. Eur. Soc. Mycobacteriol., Predominant tuberculosis spoligotypes, Delhi, India. Emerg. Infect. Dis. Istanbul, Turkey. 10:1138–1142. 18. Mathuria, J. P., P. Sharma, P. Prakash, J. K. Samaria, V. M. Katoch, and 31. Supply, P., S. Lesjean, E. Savine, K. Kremer, D. van Soolingen, and C. S. Anupurba. 2008. Role of spoligotyping and IS6110-RFLP in assessing Locht. 2001. Automated high-throughput genotyping for study of global genetic diversity of Mycobacterium tuberculosis in India. Infect. Genet. Evol. epidemiology of Mycobacterium tuberculosis based on mycobacterial inter- 8:346–351. spersed repetitive units. J. Clin. Microbiol. 39:3563–3571. 19. Mazars, E., S. Lesjean, A. L. Banuls, M. Gilbert, V. Vincent, B. Gicquel, M. 32. Tsolaki, A. G., S. Gagneux, A. S. Pym, Y. O. Goguet de la Salmoniere, B. N. Tibayrenc, C. Locht, and P. Supply. 2001. High-resolution minisatellite- Kreiswirth, D. Van Soolingen, and P. M. Small. 2005. Genomic deletions based typing as a portable approach to global analysis of Mycobacterium classify the Beijing/W strains as a distinct genetic lineage of Mycobacterium tuberculosis molecular epidemiology. Proc. Natl. Acad. Sci. U. S. A. 98:1901– tuberculosis. J. Clin. Microbiol. 43:3185–3191. 1906. 33. Valcheva, V., I. Mokrousov, O. Narvskaya, N. Rastogi, and N. Markova. 20. McGrath, J. W. 1988. Social networks of disease spread in the lower Illinois 2008. Utility of new 24-locus variable-number tandem-repeat typing for valley: a simulation approach. Am. J. Phys. Anthropol. 77:483–496. discriminating Mycobacterium tuberculosis clinical isolates collected in Bul- 21. Meyers, L. A., B. R. Levin, A. R. Richardson, and I. Stojiljkovic. 2003. garia. J. Clin. Microbiol. 46:3005–3011. Epidemiology, hypermutation, within-host evolution and the virulence of 34. van Crevel, R., E. Karyadi, F. Preyers, M. Leenders, B. J. Kullberg, R. H. Neisseria meningitidis. Proc. Biol. Sci. 270:1667–1677. Nelwan, and J. W. van der Meer. 2000. Increased production of interleukin 22. Mistry, N. F., A. M. Iyer, D. T. D’Souza, G. M. Taylor, D. B. Young, and N. H. 4 by CD4 and CD8 T cells from patients with tuberculosis is related to Antia. 2002. Spoligotyping of Mycobacterium tuberculosis isolates from mul- the presence of pulmonary cavities. J. Infect. Dis. 181:1194–1197. tiple-drug-resistant tuberculosis patients from Bombay, India. J. Clin. Mi- 35. Vitol, I., J. Driscoll, B. Kreiswirth, N. Kurepina, and K. P. Bennett. 2006. crobiol. 40:2677–2680. Identifying Mycobacterium tuberculosis complex strain families using spoli- 23. Mokrousov, I., W. W. Jiao, G. Z. Sun, J. W. Liu, V. Valcheva, M. Li, O. gotypes. Infect. Genet. Evol. 6:491–504. Narvskaya, and A. D. Shen. 2006. Evolution of drug resistance in different 36. World Health Organization. 2008. Global tuberculosis control—surveil- sublineages of Mycobacterium tuberculosis Beijing genotype. Antimicrob. lance, planning, ，nancing. WHO/HTM/TB/2008.393. World Health Organi- Agents Chemother. 50:2820–2823. zation, Geneva, Switzerland. 24. Newton, S. M., R. J. Smith, K. A. Wilkinson, M. P. Nicol, N. J. Garton, K. J. Staples, G. R. Stewart, J. R. Wain, A. R. Martineau, S. Fandrich, T. Smallie, 37. World Health Organization. 2000. Global project on anti-tuberculosis drug B. Foxwell, A. Al-Obaidi, J. Sha；, K. Rajakumar, B. Kampmann, P. W. resistance in the world. Anti-tuberculosis drug resistance in the world. Re- Andrew, L. Ziegler-Heitbrock, M. R. Barer, and R. J. Wilkinson. 2006. A port no. 2: prevalence and trends. WHO/CDS/TB/2000.278. World Health deletion de，ning a common Asian lineage of Mycobacterium tuberculosis Organization, Geneva, Switzerland.