From 2000 to 2005, positive cultures were obtained from 13 patients with a diagnosis of M. bovis infection. The strains were screened by using pncA gene for resistance to pyrazinamide sequencing, and all displayed the 169 C→G mutation (7). To differentiate between M. bovis and M. bovis BCG, we tested for the presence or absence of the region of difference 1 because the absence of this region is a specific marker of BCG strains (7,8). Spoligotyping was performed in accordance with Kamerbeek guidelines, and the data were compared with the Institute Pasteur (IP) Spoligotype Database and with the International M. bovis Spoligotype Database (9,10). We performed MIRU-VNTR typing as described by Supply et al. (11,12).

Patient age, sex, sample site, and country of birth are provided in Table 1. Most of the clinical samples were from lymph nodes (n = 6). Others samples were from urine (n = 2), lung (n = 1), sputum (n = 1), cerebrospinal fluid (n = 1), ascitic fluid (n = 1), and synovial fluid (n = 1). Patient SO, who was 4 years old when his condition was diagnosed, had been born in France, but he spent months in Algeria with his grandmother who was ill with TB. Patient GD had a history of BCG-disseminated infection after being vaccinated with a BCG strain when he was 1 year of age. His condition had also been diagnosed as a familial form of septic granulomatosis, and he was immunocompromised. The strain was isolated only after he underwent lymph node resection at the age of 17. The bacillus isolated was an M. bovis BCG strain. Patient BL had undergone immunotherapy with a BCG strain for bladder cancer, and a BCG infection of the bladder developed.

The results of spoligotyping and MIRU are shown in Table 2. Spoligotype profiles were typical of M. bovis with the absence of spacers 3, 9, 16, and 39–43 (1,13). Four distinct patterns were identified; the main one corresponded to spoligotype 482 in the IP database (70% of strains); both BCG strains exhibited this pattern. Others patterns represented were spoligotype 481 (2 strains) and 2 that were not included in the IP database (although one was identified as SB0914 in the international spoligotype database). These 2 spoligotypes (481 and 482) have been reported by Haddad et al. as the ones most commonly seen in bovine TB in France (1). Patient MB's spoligotype was not found in the databases, likely because of its origin (this patient was born in Djibouti), and it could be native to Africa.

MIRU typing identified 12 individual patterns; 2 strains possessed the same MIRU patterns but not the same spoligotype. Both BCG strains showed the same pattern, except at locus 4 (14). Patient BL was found to have a BCG strain with 1 copy on locus 4. This profile is very similar to that of the Connaught strain used for the treatment of bladder cancer, which also has 1 copy at locus 4. Patient GD's strain of BCG had 2 copies at locus 4. This characteristic is similar to that of the BCG strain used for human vaccination in France (Mérieux strain derived from the Glaxo 1077 strain) (14).

This 5-year study of human M. bovis infections in humans leads to 3 main conclusions. First, we observed a relatively high incidence of this disease: 2% of TB cases were caused by M. bovis, compared with 0.5% reported 10 years earlier and ≈1% reported in England by Smith in 2004 (15). Second, in France TB caused by MTBC occurs mainly in patients born abroad (55%), whereas in this study 70% of TB due to M. bovis occurred in French-born patients (4 of the patients had been born abroad). Therefore, human disease due to M . bovis, in contrast with that due to M. tuberculosis, appears to be predominantly indigenous in France, according to our study. However, we must note that human M. bovis infection varies throughout the world, even in industrialized countries, as reported in MMWR in 2005 when patients infected in New York were young persons born in Mexico or children of Mexican-born parents (16). Finally, we should note that French patients with M. bovis infections, in contrast to patients born abroad, were usually >50 years of age and sought treatment for a torpid infection. Measures to reduce bovine TB and the human transmission of M. bovis began in the 1950s. The disease was due to the reactivation of a past infection that had been acquired before milk pasteurization rather than a primary infection. Few cases have been reported in French-born children, which is in accordance with the effectiveness of preventive measures and their long-term effect. We cannot tell whether this is an emerging or a reemerging disease, but M. bovis is clearly still responsible for human TB. Global monitoring is required to confirm the progress of the disease and perhaps to explain why it is (re)emerging. In summary, we found the combination of spoligotyping and MIRU-VNTR to be a useful tool for identifying M. bovis infections and for determining whether patients were infected with the same strain. In our population of patients in Lyon, France, we did not detect any clonal epidemiologic features for M. bovis disease.