Mycobacterium bovis Strains Causing Smear-Positive Human Tuberculosis, Southwest Ireland

Mycobacterium bovis caused 3% of human tuberculosis cases in southwest Ireland during 1998–2006. Of 11 M. bovis strains genotyped, 9 belonged to common animal spoligotypes. Seven strains were from sputum and potential sources of human-centered disease transmission. Ten-locus variable-number tandem repeat typing gave unique strain profiles and would detect disease outbreaks.

a molecular epidemiology study that used spoligotyping, mycobacterial interspersed repetitive units-variable-number tandem repeat (MIRU-VNTR) typing, and region of difference (RD) typing of M. bovis strains isolated from human residents of Ireland.

The Study
During 1998-2006, the microbiology laboratory at Cork University Hospital obtained M. tuberculosis complex isolates from 501 patients (equivalent to 68.5% of notifi ed cases) residing in southwest Ireland (counties Cork and Kerry); 15 were M. bovis isolates (3%). Eleven of these M. bovis strains were available for testing. Seven additional isolates obtained over this period (from inoculation abscesses) were identifi ed as M. bovis BCG and not analyzed. Strains were identifi ed as M. tuberculosis complex by using Accuprobe (Gene-Probe, San Diego, CA, USA) and as potential M. bovis strains by pyruvate dependence. Defi nitive identifi cation was performed at the Mycobacterial Reference Unit in London and was based on absence of niacin production and nitrate reductase activity, thiophen-2-carboxylic acid hydrazide negativity, and pyrazinamide resistance (1). DNA extraction was performed as described (2). DNA controls were M. tuberculosis H37Rv and M. bovis AF2122/97. A 6-locus panel VNTR (exact tandem repeat [ETR]-A to ETR-F) (7) was used initially, then a 10locus VNTR panel was used (Table 1) (8). Spoligotyping was as described by Kamerbeek (9). RD analysis was conducted by the method of Brosch et al. (10) for RD1, RD4, RD9, RD10, RD12, RDpan, RD17, N-RD17, and N-RD25. This study was reviewed and approved by the Clinical Research Ethics Committee of the Cork Teaching Hospitals (Ref ECM 3).
Spoligotype signatures were recorded as binary numbers (1 and 0 denoting presence and absence of oligonucleotide spacer, respectively). Signatures were matched against the M. bovis spoligotype database (www.mbovis. org) and the global spoligotype database, SpolDB4. Unmatched spoligotypes were sent to the M. bovis spoligotype database curator for an authoritative name assignment (SB number). The discriminatory power of VNTR and spoligotyping was calculated by using the Hunter Gaston Discriminatory Index. Two or more spoligotype or VNTR patterns with 100% identity were considered a cluster.
Eight spoligotypes ( Table 2) were identifi ed among 11 isolates. Two clusters were identifi ed: 3 strains of SB0140 (ST683) and 2 strains corresponding to SB0139 (ST680) ( Table 2). SB0140 (also known as spoligotype A1 or ST1), is the most common spoligotype in animals in Ireland and the United Kingdom (1,11) ( Table 2). SB0139 was previously detected as an isolate from a cow in Northern Ireland in 2000 (SpolDB4; R. Skuce, pers. comm.). Other spoligo-types we identifi ed that had been previously reported in isolates from animals in Ireland were SB0130 (ST691), also described as D1 or ST7; SB0142 (ST679), also described as A2 or ST2; and SB1047 (not found in SpolDB4). Three strains were not found in either of the databases. These strains were assigned spoligotypes SB1185, SB1186, and SB1187 through the M. bovis spoligotype database. SB1186 and SB1187 differed from the ubiquitous SB140 by absence of spacer 35 and spacers 35 and 37, respectively. Previously reported SB0139, SB0142, and SB1047 differed from SB0140 by absence of spacers 33, 33-34, and 35-36, respectively. SB1185 differed from the established animal strain SB0130 by the absence of spacers 20-21.

Conclusions
A recent outbreak report describing sputum-positive M. bovis disease transmitted by person-to-person contact (6) underlines the need for precise genetic markers of M. bovis to aid epidemiologic traceback. We studied strains of M. bovis isolated from humans in the Republic of Ireland, and we have defi ned an optimal set of markers using a combination of spoligotyping and VNTR. We detected a group of isolates ( Table 2) of spoligotype SB0140, which is predominant in animal strains of M. bovis reported from the Republic of Ireland (51.8% of isolates) (11) and the United Kingdom (1). It forms the single largest group of M. bovis strains isolated from humans in the United Kingdom (30%) (1). This group was not reported in a recent series of M. bovis isolates from humans in Italy (13) or France (14). Indeed, none of the spoligotypes in our survey and these reports overlap. Predominant strains by spoligotype in animals and those infecting humans in the same country are known to be linked (1,13,14). We found 3 novel spoligotypes similar to SB0140 in a small group of patients, showing that a wider variety of strains infect humans than animals, as described in similar studies (1,2,13).
In our study, 81% of patients infected were >30 years of age (Table 2), comparable with fi ndings of a previous survey of the southwest Ireland population (5). Primary infection of this group is likely to have been several decades before diagnosis, and our isolates probably represent reactivation of disease acquired earlier in life, effectively a record of past prevalence in animals. Spoligotype SB0140 strains were isolated from 2 patients who were >80 years of age, showing that the current predominance of SB0140 in animals (11) is therefore of long duration in Ireland, potentially going back 8 decades.
Molecular typing by insertion sequence (IS) 6110based restriction fragment length polymorphism is inadequate for M. bovis because of low copy numbers of IS6110. A combination of MIRU-VNTR and spoligotyping gives better discrimination (15). Traditional 6-locus VNTR (ETR-A-ETR-F) has been described for typing of M. tuberculosis complex strains (7) including M. bovis (2), but an expanded panel with an additional 10 loci applied to our strains greatly improved discrimination and enabled individual identifi cation of each isolate.
Seven of our 11 isolates were from sputum, and 6 were detected on direct smear with potential for transmission. A VNTR-typing scheme based on the loci established on M. bovis isolates from animals (8) would detect M. bovis clusters derived from foodborne outbreaks or horizontal trans-mission of disease between humans in Ireland. Our study provides ways to markedly improve the ability to identify and contact-trace future clusters of M. bovis infection in humans.