Phenotypic and genotypic traits and epidemiological implication of Vibrio cholerae O1 and O139 strains in India during 2003

Vibrio cholerae strains belonging to serogroups O1 and O139 are responsible for the vast majority of epidemic cholera in humans (Bennish, 1994), which continues to be a growing concern in most developing countries. V. cholerae O1 is further divided into two biotypes, Classical, and El Tor and two major serotypes, Inaba and Ogawa. Since its emergence in 1961 as a causative agent of the seventh cholera pandemic, V. cholerae O1 El Tor has spread to almost 75 countries in Asia, Africa and South America. In 2005, a sharp increase of cholera cases with increased mortality was reported from 52 countries, and identified 131 943 cases with 2272 deaths (World Health Organization) (WHO, 2006). The changing trends in the epidemiology of pathogenic V. cholerae has presented problems for the control of the disease. A good example is the sudden emergence of V. cholerae O139 strains in the Indian subcontinent during late 1992 and its rapid spread to neighbouring countries (Ramamurthy et al., 1993). The clinical picture of diarrhoea caused by either V. cholerae O1 or O139 strains is indistinguishable and the similar propensity to spread lead both to be classified as the agents of epidemic cholera (Bhattacharya et al., 1993). It has also been suggested that V. cholerae O139 strains possibly evolved from the seventh pandemic clone of V. cholerae O1 El Tor (Berche et al., 1994; Popovic et al., 1995).

The last decade has witnessed periodic switching between V. cholerae O1 and O139 strains as agents of cholera in India and Bangladesh (Basu et al., 2000; Faruque et al., 2003). Extensive use has been made of molecular tools to index changes occurring among circulating strains, and a number of ribotypes and DNA macrorestriction pattern types have been identified in strains of both serotypes (Dutta et al., 2006; Faruque et al., 2000; Garg et al., 2000; Nair et al., 2002; Popovic et al., 1993; Safa et al., 2005; Sharma et al., 1997; Yamasaki et al., 1997).

During 2004 in many parts of India, the re-emergence and progression of V. cholerae O1 serotype Inaba causing outbreaks and sporadic infections was documented by Taneja et al. (2005). This prompted us to analyse V. cholerae O1 and O139 strains isolated from acute diarrhoeal patients admitted to hospitals in six different parts of India prior to the spread of the Inaba serotype.

Study design and bacteriology. V. cholerae strains isolated from cholera patients admitted to different hospitals during 2003 were selected from the archive collection of the National Institute of Cholera and Enteric Diseases, which is the national reference centre. During 2003, 247 clinical strains were identified as V. cholerae and comprised 133 strains from patients in different parts of India [37 from Chennai, 4 from Manipal (southern India), 9 from Mumbai, 21 from Ahmedabad (western India), 10 from Raygaunge (eastern India), 34 from Aligarh and 18 from Haryana (northern India)] and 114 strains from patients admitted to the Infectious Diseases Hospital, Kolkata.

Stools or rectal swabs were plated on thiosulfate/citrate/bile salts/sucrose agar. Presumptive V. cholerae isolates formed sucrose-fermenting yellow-pigmented colonies, which were oxidase-positive and produced an alkaline slant and acid butt (K/A) in Kapers multitest medium (Nair et al., 1987). V. cholerae strains were typed using antisera specific for O1 and O139 serogroups that were raised in our institute. Strains of O1 serogroup were further classified with monospecific antisera to the Ogawa and Inaba serotypes.

Antibiotic susceptibility assay. The susceptibility of strains to 11 antimicrobial agents was determined by a disc diffusion assay (Bauer et al., 1966) with commercially available discs (HiMedia) containing ampicillin (10 µg), chloramphenicol (30 µg), co-trimoxazole (25 µg), ciprofloxacin (5 µg), furazolidone (100 µg), gentamicin (10 µg), neomycin (30 µg), nalidixic acid (30 µg), norfloxacin (10 µg), streptomycin (10 µg) and tetracycline (TET) (30 µg). The inhibition zone diameters were recorded, and strains were classified as susceptible, reduced susceptible or resistant to a particular antimicrobial agent according to the published criteria (NCCLS, 2002). E. coli strain ATCC 25922 was used for quality control.

PCR and randomly amplified polymorphic DNA (RAPD) analysis. Purified genomic DNA was isolated from the test strains grown in LB medium at 37 °C overnight (Murray & Thompson, 1980). Briefly, cells were harvested, washed and resuspended in TE buffer (10 mM Tris/HCl, 1 mM EDTA) pH 8.0 and then lysed by treatment with SDS, followed by digestion with proteinase K (20 µg ml¹) for 2 h at 60 °C. Genomic DNA was further purified by treatment with 10 % N-cetyl-N,N,N-trimethylammonium bromide in 0.7 M NaCl at 65 °C for 10 min and sequential extraction with phenol/chloroform/isoamyl alcohol (25 : 24 : 1, by vol.) and chloroform. The DNA was precipitated with ethanol, air-dried, reconstituted in sterile water and stored at 4 °C. This DNA was used in the PCR assay with primers specific for ctxA and allele-specific tcpA (De et al., 2001; Keasler & Hall, 1993). In separate assays, allele-specific rstR primers were used for the detection of allelic types of rstR among different strains (Bhattacharya et al., 2006). The PCR assays were carried out in a final reaction volume of 25 µl, and the concentration of primers and dNTPs adjusted to 1 pmole µl¹ and 0.25 µM, respectively. The ctxAtcpA multiplex PCR assay consisted of initial denaturation at 94 °C for 2 min, followed by 30 cycles comprising 1.5 min each at 94, 60 and 72 °C, and a final extension at 72 °C for 7 min. The rstR allele-specific PCR assays consisted of initial denaturation at 94 °C for 2 min, followed by 30 cycles of 40 s at 94 °C, 40 s at 50 °C and 1.5 min at 72 °C, and a final extension for 7 min at 72 °C. For RAPD-PCR, about 20 ng purified genomic DNA was used in each assay with primer 1281 (Sinha et al., 2002). Cycling conditions consisted of an initial denaturation at 94 °C for 5 min, followed by 45 cycles of 1 min at 94 °C, 1 min at 36 °C and 2 min at 72 °C (Sinha et al., 2002). Amplification products were separated in agarose gels, stained with ethidium bromide, viewed under UV and recorded digitally using a gel documentation system (GelDoc 2000, Bio-Rad).

DNA probes and Southern blot hybridization. Purified genomic DNA was treated with restriction endonuclease enzymes and separated by electrophoresis in 0.8 % agarose gels. DNA fragments were denatured by treatment with alkali and subsequently transferred to nylon membrane (Hybond-N⁺; Amersham Pharmacia Biotech), according to the procedure of De et al. (2005), and hybridized with a DNA probe. Ribotyping analysis was carried out with BglI digested genomic DNA using a 7.5 kb BamHI fragment (from the plasmid pKK3535) as a probe for rrn genes (Brosius et al., 1981). For CTX typing, genomic DNA was digested individually with BglI, BglII, PstI, XbaI, MluI, HindIII and AvaI (Takara, Shuzo). A 540 bp XbaIClaI fragment of ctxA, cloned in pKTN901 using the EcoRI linker, served as a probe for ctxA (Kaper et al., 1988). The 251 bp PCR amplicon of cep was cloned into PCR 2.1 (Invitrogen) to generate clone pSC01. This clone was sequenced using M13 primers to confirm the presence of cep. Similarly, the clones pSC02, pSC06 and pSC10 were generated by the cloning of allele-specific rstR amplicons of Classical, El Tor and Calcutta types, respectively, and their identities were confirmed by nucleotide sequencing. EcoRI digested products of pSC clones served as specific probes for allele-specific rstRs and cep. DNA probes were labelled with chemiluminescent dye (Gene Images AlkPhos direct labelling and detection system; Amersham Biosciences) and hybridization reactions were developed following the manufacturers protocol and recognition patterns recorded on X-ray film.

PFGE. Agarose embedded genomic DNA of the V. cholerae strains was cleaved with 50 U NotI (Takara) and the fragments were separated in a contour-clamped homogeneous electric field (CHEF Mapper; Bio-Rad) by the procedure of Garg et al. (2000). Run conditions were generated by the autoalgorithm mode of the CHEF Mapper system using a size range of 20 to 300 kb with an electrophoresis time of 40 h, 24 min in 1 % PFGE grade agarose gel in 0.5x TBE [44.5 mM Tris/HCl, 44.5 mM boric acid, 1.0 mM EDTA (pH 8.0)] at 14 °C. The gels were stained for 30 min in Elix Milli-Q water (Millipore) containing 1.0 µg ethidium bromide ml¹, destained in Elix Milli-Q water for 15 min and photographed under UV light using the Gel Doc 2000 gel documentation system (Bio-Rad).

Nucleotide sequencing and analysis of wbeT. A 902 bp region of wbeT was amplified by PCR using primers VFBT-F and VFBT-R (Rijpkema et al., 2004). The amplified product was purified (PCR purification kit; Qiagen) and used directly as a template for nucleotide sequencing. Both strands of DNA were sequenced with the BigDye terminator cycle sequencing kit (Applied Biosystems) according to the manufacturers instructions, using an ABI PRISM 310 DNA sequencer (Applied Biosystems). The nucleotide and deduced amino acid sequences were compared to published sequences available in GenBank with DNASTAR (DNA Star).

Phenotypic traits of V. cholerae O1 and O139 strains in 2003
Serogroup analysis assigned 240 strains to O1 and 7 strains to O139 serogroups. A total of 1107 diarrhoeal specimens were collected from the Infectious Diseases Hospital, Kolkata: 112 were serogroup O1 Ogawa and 2 were serogroup O139 (Table 1). No serotype Inaba strains were found. A periodic shift between V. cholerae O1 and O139 from cholera patients was observed in India and Bangladesh (Basu et al., 2000), but despite the re-emergence of V. cholerae O139 in 2002 in Bangladesh (Faruque et al., 2003), this serogroup appears to be mostly restricted to certain areas in China (WHO, 2005, 2006). Indeed, since 2003 serogroup O139 has been rarely isolated from Kolkata and V. cholerae O1 serotype Ogawa remains predominant. The absence of the Inaba serotype accords with reported patterns of V. cholerae O1 strains in the Indian subcontinent. With the exception of the recently reported increase in its prevalence (Dutta et al., 2006), serotype Inaba was last predominant during 1989 (Ramamurthy et al., 1992), and was then isolated sporadically during 19981999 (Garg et al., 2000). Moreover, O1 Inaba strains from the latter period were reported to have evolved from the then circulating Ogawa strains (Garg et al., 2000). The fact that switching between Inaba and Ogawa serotypes was demonstrated to occur with relative ease (Manning et al., 1994), lead us to speculate whether the O1 Ogawa strains of 2003 were the precursors of the O1 Inaba strains that started to appear since mid 2004.

Table 1. Antimicrobial percentage resistance patterns exhibited by V. cholerae O1 and O139 strains from different parts of India during 2003 Apart from Kolkata and Royguange strains, strains were received from the King Institute of Preventive Medicine, Chennai; the Kasturba Medical College, Manipal; Seth G. S. Medical College & KEM Hospital, Mumbai; Sheth V. S. General Hospital, Ahmedabad; Aligarh Muslim University, Aligarh; and the Post Graduate Institute of Medical Science, Haryana. SP, Sporadic case; O, outbreak case; AMP, ampicillin; CHL, chloramphenicol; CIP, ciprofloxacin; CO, co-trimoxazole; FZ, furazolidone; GEN, gentamicin; NEO, neomycin; NAL, nalidixic acid; NOR, norfloxacxin; STR, streptomycin.

Approximately half (66) of the V. cholerae clinical isolates from different parts of India and 106 isolates from Kolkata were tested for antimicrobial susceptibility. Table 1 shows that the great majority of isolates showed resistance to ampicillin, furazolidone and nalidixic acid, and many were resistant to co-trimoxazole and streptomycin. Resistance profiles were variable and not associated with particular strains or serogroups. Interestingly, 45.7 % Ogawa strains showed reduced susceptibility to ciprofloxacin, while 91 % of Inaba strains recovered in 20042005 were also reported to be less susceptible to this drug (Dutta et al., 2006). All isolates, except one, were susceptible to chloramphenicol and gentamicin, and apart from seven isolates from Chennai and one from Aligarh, neomycin susceptibility was universal. With the exception of three serogroup O1 Ogawa strains from Kolkata, none exhibited resistance to TET. Resistance to TET among O1 Inaba strains that were isolated in India during 2005 has already been reported (Jesudason, 2006).

Virulence genes and genotypes
All strains, except one from Aligarh, showed the presence of an identical El Tor tcpA allele, and all harboured the ctxA encoding A subunit of the cholera toxin. Sixty-six V. cholerae O1 and six O139 strains were selected for molecular analysis. The serogroup O1 strains were all indistinguishable by RAPD-PCR, but the O139 strains were separated into two groups based on the presence of an additional amplification band (Fig. 1). This single difference may not be significant in terms of strain discrimination and suggests that isolates within each serogroup formed single or very closely related clones. It is noteworthy that the RAPD profile of the O1 strains resembled the reported profiles of O1 Ogawa strains that caused an outbreak of cholera in Orissa during 1999 (Chhotray et al., 2002). Furthermore, the O139 strains lacking the additional band (Fig. 1b, lanes 2, 3, 6) displayed a profile identical to that reported for O139 strains from an outbreak in Ahmedabad in 2000 (Chakraborty et al., 2001).

(99K): Fig. 1. RAPD-PCR profiles of V. cholerae strains of serogroups (a) O1 and (b) O139 with primer 1281. (a) Lanes: 1, Ahmedabad, Ahm6; 2, Royguange, WD15; 3, Kolkata, SC11; 4, Kolkata, SC213; 5, Aligarh, AMU12/03; 6, Mumbai, MUM21; 7, Haryana, HAR5B; 8, Chennai, VCH156; 9, Manipal, VCA. (b) Lanes: 1, Haryana, HAR3A; 2, Haryana, HAR6B; 3, Chennai, VCH133; 4, Chennai, VCH209; 5, Kolkata, SC185; 6, Kolkata, SC199. The size marker was 1 kb ladder.

All of 32 serogroup O1 Ogawa strains (two random strains from each region) analysed by ribotyping were identical in hybridization profile (Fig. 2) and this corresponded to pattern RIII, which was first described for O1 strains that emerged in 1993 after the spread of the O139 serogroup subsided in many cholera endemic regions (Sharma et al., 1997). The six O139 strains were also indistinguishable from each other and the profile was classified as the pattern BII (Faruque et al., 2000)</xref> (Fig. 2). However, newly emerged Inaba strains during 20042005 displayed in addition to RIII, a new ribotype pattern designated type RIV (Dutta et al., 2006). Furthermore, the ribotypes of the O1 strains circulating in India during 2003 are different from O1 strains, with attributes of both Classical and El Tor biotypes (Matlab variant) described from Bangladesh (Nair et al., 2002).

(75K): Fig. 2. BglI ribotype patterns of V. cholerae strains of serogroups (a) O1 and (b) O139. (a) Lanes: 1, Ahmedabad, Ahm6; 2, Royguange, WD15; 3, Kolkata, SC11; 4, Kolkata, SC213; 5, Aligarh, AMU12/03; 6, Mumbai, MUM21; 7, Haryana, HAR5B; 8, Chennai, VCH156; 9, Manipal, VCA. (b) Lanes: 1, Chennai, VCH133; 2, Chennai, VCH209; 3, Haryana, HAR3A; 4, Haryana, HAR6B; 5, Kolkata, SC185; 6, Kolkata, SC199. The size marker was HindIII-digested λ DNA.

Segregation of multiple clones of V. cholerae O1 and O139
PFGE analysis has been widely used for analysing the clonal nature of several groups of bacteria, and has been applied for the distinction of multiple clones of V. cholerae that have emerged and declined in India and Bangladesh (Bag et al., 1998; Cooper et al., 2006; Safa et al., 2005; Yamasaki et al., 1997). Indeed, six new pulsotypes, in addition to the five established types, have been reported recently for V. cholerae O1 strains that appeared during 20042005 in India (Dutta et al., 2006). In this study, PFGE of V. cholerae O1 Ogawa strains from different parts of India were grouped into two types, coded types H (25 strains; Fig. 3a, lanes 1, 3, 8, 9) and L (8 strains; Fig. 3a, lanes 2, 47). The latter represents a new pulsotype first detected in O1 Inaba strains during 20042005 (Dutta et al., 2006), but pulsotype H has prevailed among O1 strains for the last decade. The predominant pulsotype pattern of the O139 strains was comparable to that of strains isolated during 2000 from an outbreak in Ahmedabad (Fig. 3b, lanes 1, 2, 5, 6) (Chakraborty et al., 2001). Two new pulsotype profiles were given by two strains isolated from Haryana, but the widely distributed pulsotypes F and G described by Yamasaki et al. (1997) were not found, suggesting that a new clone of O139 may have already displaced older pulsotype F and G clones.

(110K): Fig. 3. NotI PFGE profiles of V. cholerae strains of serogroups (a) O1 and (b) O139. (a) Lanes: 1, Ahmedabad, Ahm6; 2, Royguange, WD15; 3, Kolkata, SC11; 4, Kolkata, SC213; 5, Aligarh, AMU12/03; 6, Mumbai, MUM21; 7, Haryana, HAR5B; 8, Chennai, VCH156; 9, Manipal, VCA. (b) Lanes: 1, Chennai, VCH133; 2, Chennai, VCH209; 3, Haryana, HAR3A; 4, Haryana, HAR6B; 5, Kolkata, SC185; 6, Kolkata, SC199. The size marker was λ ladder.

The copy number and organization of the CTX genetic element (CTX prophage) in V. cholerae strains was investigated. It has already been established that toxigenic traits of V. cholerae strains are due to lysogenic conversion of CTXφ carrying ctxAB (Waldor & Mekalanos, 1996). Interestingly, the genes encoding cholera toxin (ctxAB) were also shown to be acquired by the pre-CTXφ, which was originally devoid of ctxAB. Furthermore, detailed studies established that primarily three allelic types of CTX prophages (chromosomally integrated form of CTXφ) exist in toxigenic V. cholerae O1 and O139 (Davis et al., 1999; Kimsey & Waldor, 1998). The nucleotide sequence variations in the repressor region rstR formed the basis of the distinct alleles, namely CTX^Cl, CTX^El and CTX^Cal (Davis et al., 1999; Kimsey & Waldor, 1998). Initially, CTX^Cl and CTX^El types were detected in O1 strains of Classical and El Tor biotypes, respectively, and accordingly named, and CTX^Cal and CTX^El also have been identified in O139 strains (Davis et al., 1999). Nevertheless, Matlab variants with attributes of both biotypes have caused cholera in both Bangladesh (Nair et al., 2002) and Mozambique (Lee et al., 2006). We screened the V. cholerae O1 strains by PCR assays with allele-specific rstR (Bhattacharya et al., 2006) and found that, with the exception of one strain from Chennai (VCH18), all strains produced amplicons specific to El Tor type of rstR (Fig. 4). It is evident from Fig. 4(b, c) that strain VCH18 (lane 1) possessed both the El Tor as well as the Cal type of rstR, and this may have been acquired by lysogenic conversion of CTXφ^Cal into the O1 strain. This is an unusual finding as lysogenic conversion of CTXφ^Cal had been reported for O139 V. cholerae strains (Davis et al., 1999). PCR-based analysis revealed all six O139 strains had both the El Tor and the Cal type of rstR (data not shown). These strains might therefore have acted as a precursor of CTXφ^Cal that lysogenized into certain O1 strains in the same region during the study period.

(42K): Fig. 4. PCR amplicon profiles of V. cholerae strains obtained with primers specific to (a) rstR of Classical biotype (rstRCl), (b) El Tor biotype (rstREl) and (c) Calcutta biotype (rstRCal). Lanes: 1, Chennai, VCH18; 2, Aligarh, AMU22/03; 3, Royguange, WD15; 4, Manipal, VCA; 5, Kolkata, SC32; 6, reference strain V. cholerae O1 Classical, O395; 7, El Tor, CO366; 8, O139, AS207. Amplicon sizes of 474, 501 and 313 bp specific for rstRCl, rstREl and rstRCal, respectively, are indicated.

To determine the copy number of the CTX genetic element, DNA from V. cholerae strains was digested with different restriction enzymes with unique recognition sites within the CTX element. Southern blot hybridization with a ctxA probe revealed the presence of one copy of CTX in serogroup O1 Ogawa strains, except for VCH18. The complete organization of CTX genetic elements in the strain VCH18 could not be ascertained and further studies on this aspect are in progress. On the other hand, DNA from O139 strains produced multiple bands on digestion with HindIII and separate probing with ctxA, rstR^Cal and cep (Fig. 5); the results generated with different enzymes are summarized in Table 2. Based on these data and allele-specific rstR, the organization of copies of the CTX genetic element in O139 strains was postulated as presented in Fig. 6. The detection of a single DNA fragment, with an estimated size of 7.2 kb, generated in the ctxA RFLP assays using most of the enzymes (except for HindIII) can be explained by the presence of one additional copy of CTX prophage devoid of ctxAB (truncated CTX prophage) downstream to that of two tandemly arranged CTX^El and CTX^Cal prophages (Fig. 6b). The predicted three copies of the prophages were further confirmed by hybridization analysis with cep and allele-specific probes. The O139 strain isolated from Chennai (VCH209) displayed an altered profile as compared to the other O139 strains. Studies on the organization of the CTX genetic element in VCH209 are under way. The finding of a new CTX prophage profile in most of the O139 strains to the best of our knowledge is novel and has not been reported before.

(14K): Fig. 5. Southern blot hybridization patterns of HindIII-digested genomic DNA of V. cholerae O139 strains with DNA probes specific to (a) ctxA, (b) rstRCal and (c) cep. Lanes: 1, Haryana, HAR3A; 2, Haryana, HAR6B; 3, Chennai, VCH133; 4, Chennai, VCH209, 5, Kolkata, SC185; 6, Kolkata, SC199; 7, reference V. cholerae O139 strain AS207. The size marker was HindIII-digested λ DNA.

Table 2. Southern hybridization patterns generated with probes specific to ctxA and rstR alleles among V. cholerae O139 strains isolated during 2003 in India

(25K): Fig. 6. Schematic diagram showing the genomic organization of CTX prophages (not to scale) in (a) V. cholerae O1 and (b) O139 strains, as predicted from the data obtained in the Southern blot hybridization assay using DNA probes specific to ctxA, allele-specific rstR and cep. ctxAB, RS1 and RS2, comprising either rstREl or rstRCal, are indicated. The DNA probe and corresponding hybridized fragments generated by restriction enzymes are shown. A, AvaI; BI, BglI; BII, BglII; H, HindIII; P, PstI.

The analysis was extended further by determining the nucleotide sequence of the wbeT region of the O1 Ogawa strains. Sequence data revealed complete identity of the wbeT region to that of the O1 Ogawa strains (GenBank accession nos AY458622, AY458623, AY458624), but the region differed from those of the previously isolated O1 Inaba strains of 1989 and 19981999 in India (GenBank accession nos AY458621, AY458625, AY458626) (Garg et al., 2000; Rijpkema et al., 2004). However, comparative analysis revealed wbeT of 2003 O1 Ogawa differed from that of the recently emerged Inaba by a single point mutation, and that mutation has already been reported as unique for the newly emerged Inaba serotype.

In conclusion, the phenotypic and the genotypic analysis of V. cholerae strains isolated in 2003 has an added epidemiological importance considering the emergence and spread of the Inaba serotype in 2004, which gradually replaced the Ogawa serotype in several cholera endemic regions of India (Dutta et al., 2006; Taneja et al., 2005). The O1 and O139 strains studied here exhibited resistance to multiple drugs in keeping with previous reports, but it is noteworthy that except for a few strains from Kolkata none were TET resistant, although reduced susceptibilities to this drug were observed, which might result in spontaneous selection of TET resistance. A comprehensive molecular analysis gives support to the hypothesis that O1 Ogawa strains of 2003 may have been precursors for recently emerged epidemic Inaba strains. Indeed, the nucleotide sequence of wbeT was identical except for a unique point mutation reported earlier for the newly emerged Inaba strain (GenBank accession no. DQ401028) (Dutta et al., 2006); this mutation might play a role in the switching of Ogawa to Inaba serotype. Serogroup O139 strains have continued to be uncommon in India since 2003 to the present date. The continued prevalence of serogroup O1 for many years might act as a selective pressure for O139 strains to undergo slight genetic changes to ensure that they remain a significant agent in the cause of cholera. Interestingly, the O139 strains harboured two copies of complete CTX prophages, along with tandemly arranged truncated prophage characterized by an absence of ctxAB. Considering the epidemic potential of O1 and O139 strains, it is imperative to maintain molecular epidemiological surveillance of cholera infections to document subtle evolutionary changes in the constitution of V. cholerae.

We acknowledge the following for sending V. cholerae isolates: the King Institute of Preventive Medicine, Chennai; the Kasturba Medical College, Manipal; Seth G. S. Medical College & KEM Hospital, Mumbai; Sheth V. S. General Hospital, Ahmedabad; Aligarh Muslim University, Aligarh; and the Post Graduate Institute of Medical Science, Haryana. This work was supported, in part, by the Council of Scientific and Industrial Research (CSIR), the Government of India (nos 37/1206/04/EMR II) and the Japan International Cooperation Agency (JICA/NICED project 054-1061-E-O). K. G. is the recipient of research fellowships from University Grants Commission, Government of India.