Introduction

Tuberculosis (TB) remains a major global health threat and is the single most frequent cause of death by an infectious agent [1, 2]. In addition to pulmonary manifestations, extrapulmonary (EP) TB involvement has become an significant clinical problem of major concern, especially among HIV co-infected cases [3–6]. Members of the M. tuberculosis complex, which comprise the very closely related species M. tuberculosis, M. bovis, M. africanum, M. microti, M. canetti, and M. caprae [7], are the most frequent cause of the disease. M. tuberculosis is by far the most important; however, M. bovis, the agent of bovine TB, may still be considered a potential cause of human cases, especially in developing countries where control measures for bovine TB in cattle and/or milk and dairy products are not always satisfactory [8]. The advent of the human immunodeficiency virus (HIV) acquired immunodeficiency syndrome pandemic has become an important additional risk in the development of M. bovis-mediated disease [9]. In Mexico, the frequency of M. bovis-mediated diseases has decreased after the introduction of control measures of dairy products [10]. However, it is important to emphasize that EP or pulmonary M. bovis disease is very likely to be underestimated because it is clinically indistinguishable from disease caused by M. tuberculosis [11], and its laboratory diagnosis requires isolation facilities and specific identification of the bacteria. Even where mycobacterial cultures are available, M. bovis may be misdiagnosed as M. tuberculosis due to a lack of interest in the typing of mycobacteria isolates [12] or in the absence of accurate M. tuberculosis complex speciation [13]. Therefore, demonstration of M. bovis as an etiological agent in any patient with suspected TB should be compulsory due to its well-known intrinsic resistance to pyrazinamide (PZA) [14]. Appropriate and timely treatment must be implemented.

In this study we describe the prevalence of M. bovis defined by spoligotyping as the cause of EPTB, as well as the response to chemotherapeutic measures, in a group of adult HIV-positive (HIV+) and HIV-negative (HIV−) patients studied at the Pneumonology Service, Hospital General de Mexico (HGM), Mexico City from January 2000 to December 2003.

Materials and methods

A prospective study to investigate the role of M. bovis as an etiological agent for EPTB was undertaken at the Pneumonology Service of the HGM. The institutional review board approved the study protocol. HGM is an 1,100-bed tertiary-level general hospital with teaching facilities. It is a primary and reference health center mainly for low income and unemployed patients from Mexico City, the surrounding metropolitan area, and the neighboring Mexican states, with an estimated beneficiary population of ∼20 million inhabitants. As part of the routine care, all adult patients with suspected TB are tested for HIV infection, but only with the patient’s consent. Similarly, all new adult HIV+ cases (determined by enzyme immunoassay and positive Western blot results) are routinely referred to the Pneumonology Service for TB detection and other respiratory tract infections. At the time of hospital arrival, a complete clinical record was obtained and a questionnaire was filled in to record the demographic data and socio-economic status of every patient, followed by skin testing with purified protein derivative (PPD-RT23, 5TU; Statens Serum Institut, Copenhagen, Denmark).

Initial diagnoses of pulmonary TB were made on the basis of clinical findings, including cough and expectoration lasting longer than 2 weeks. In these patients, three sputum samples were collected, and acid-fast bacilli identification on smears stained using the Ziehl-Neelsen technique and culture on Löwenstein-Jensen solid slants were performed [15]. EP cases were initially identified on the basis of clinical and radiographic findings; confirmation of TB/mycobacterial etiology and organ involvement was established by positive culture. Samples of urine, cerebrospinal, pleural or peritoneal fluids, and tissue biopsies and aspirates were evaluated as recommended [16]. Histopathologic examination of tissue biopsies and aspirates was performed as described elsewhere [17].

Mycobacterium species of all patients’ isolates were identified through their distinctive molecular characteristics using the polymerase chain reaction of the heat-shock protein 65 gene (hsp65) followed by BstEII-HaeIII restriction fragment-length polymorphism analysis, as described by Devallois et al. [18]. This method allows the differentiation of the M. tuberculosis complex species as a whole; therefore, to distinguish M. bovis from M. tuberculosis isolates, spacer oligonucleotide typing (spoligotyping) analysis [19] was used. Spoligotyping was performed using a commercially available kit in accordance with the manufacturer’s recommendations (Isogen Bioscience BV, Bilthoven, The Netherlands). As a control group, a randomly selected sample of adult HIV− patients with EPTB confirmed through the isolation of mycobacteria was included.

In M. tuberculosis complex isolates, drug susceptibility was determined using the BACTEC Mycobacterial Growth Indicator Tube test 960 (MGIT; Becton-Dickinson, San Jose, CA, USA). Due to technical difficulties in obtaining growth in the very acidic medium required for PZA activity, the PZA susceptibility test for this drug was not performed routinely at HGM at the time of this study.

Immediately after bacteriological, histological or molecular confirmation of mycobacterial infection, initial treatment, consisting of a four-drug regimen (isoniazid, rifampin, ethambutol, and pyrazinamide) in a directly observed therapy-based (DOT) approach, was instigated [20]. As M. bovis is resistant to PZA [14], as soon as its definitive identification was obtained, this drug was suspended and substituted by streptomycin in a dose of 1 g i.m. for 5 consecutive days/week, and the regimen was extended up to 9 months. No further attempt to confirm PZA resistance by in vitro testing was intended. In HIV+ cases, antiretroviral therapy [21] was delayed for 4–8 weeks after starting antituberculosis drug therapy, unless a low CD4 count (<50 cells/mm3) was present, and combinations of zidovudine and lamivudine with efavirenz or nevirapine were used [22]. A follow-up of at least 6 months was planned to assess the course of both infections.

A univariate statistical analysis using Mantel and Haenzel’s Chi squared test was performed to compare differences in sex, HIV status (positive/negative), and site of infection in M tuberculosis vs M. bovis cases. A statistical analysis using a t test was used to compare differences in means of age in M tuberculosis vs M bovis cases. In both cases, p ≤ 0.05 was considered statistically significant.

Results

During the 4-year study period, there were 1,374 culture-confirmed cases of mycobacterial disease in the HGM. Of these cases, 842 patients (61%) had only pulmonary involvement and 532 (39%) had EP manifestations. In the latter group, 452 (85%) were HIV− and 80 (15%) were HIV+, all of whom had not had antiretroviral therapy before hospital admission. For this study, the 80 co-infected patients with EP involvement were included, 49 were men (61%, mean age 37 years, range 19–65 years), and 31 women (39%, mean age 40 years, range 19–72 years). As a control group, 83 patients with EP disease and negative results from HIV testing carried out with their prior consent were randomly chosen: 46 men (55%, mean age 46 years, range 22–76 years), and 37 women (45%, mean age 45 years, range 23–76 years).

A comparison of the characteristics of the study participants in relation to etiological agent and HIV status is shown in Table 1. No statistical significance was found among demographic and clinical data. The small sample size of M. bovis cases determined the outcome of this statistical analysis. The overall results showed that M. tuberculosis was the most frequent mycobacteria isolated, regardless of HIV status, and was present in 67 individuals (80.7%) in the HIV− population and in 46 patients (57.5%) in the HIV+ cases. Non-tuberculosis mycobacteria (M. avium complex, M. gordonae, and M. fortuitum) were isolated in 10 HIV-free cases (12%) and in 23 HIV+ patients (28.75%).

Table 1 Comparison of demographic and clinical characteristics of extrapulmonary tuberculosis patients according to etiological agent
Full size table

M. bovis, identified by its characteristic spoligotype consisting of deleted direct repeat spacers 39–43, was present in 6 HIV− patients (7.2%; 4 males and 2 females, mean age 39 years, range 24–65 years), and in 11 HIV-seropositive cases (13.75%; 9 males and 2 females, mean age 39 years, range 19–64 years). Table 2 shows the demographic, clinical, and outcome features of the HIV− and HIV+ cases where M. bovis was isolated. PPD reactivity in the HIV+ group of patients (cut-off level ≥5 mm) was 1/10, lower than that reported for M. tuberculosis-infected HIV+ cases [23]. EP involvement in the cases in the HIV− group consisted of 3 patients with lymphadenitis and 3 with genitourinary infection; in contrast, in the HIV+ cases 7 were associated with lymphadenitis, 3 with genitourinary infection, and 1 with meningeal infection. Figure 1 displays the M. bovis spoligotype profiles obtained; no clustering among isolates was found and every isolate displayed a unique spoligotype. When these spoligotype profiles were compared with the prototypes reported by Brudey et al. [24], all but one corresponded to the BOV1 lineage, the BOV2 lineage being the other. Interestingly, one isolate showed a profile identical to the spoligotype of the vaccine M. bovis Bacillus Calmette-Guérin (BCG) strain.

Table 2 Characteristics of the Mycobacterium bovis-infected patients
Full size table
Fig. 1
figure 1

Schematic representation of the 17 Mycobacterium bovis isolates obtained in this study. The black rectangles represent positivehybridization signals and the white rectangles depict lack of hybridization

Full size image

In the treatment of M. bovis cases, the standard regimen was extended up to 9 months after this agent was finally identified [25], with the inclusion of streptomycin instead of PZA because of the high sensitivity to the drug demonstrated by the isolates. No multidrug-resistant (MDR) isolates were found. At the end of the follow-up period, a favorable response to specific treatment was observed in the 6 HIV− patients. In the group of cases co-infected with HIV, a combination of anti-M. bovis treatment and highly active antiretroviral therapy (HAART) resulted in clinical improvement in 9 patients, without the presence of paradoxical responses; 1 patient was lost after hospitalization due to a change of residence, and 1 patient died.

Discussion

In this study, Mexican adult patients affected by M. bovis-mediated EPTB were found in a higher proportion than expected, either in immune-uncompromised or in HIV-infected patients. Confident and consistent identification of M. bovis was obtained by spoligotyping, showing a wide diversity of genetic types with no isolates sharing the same spoligotype. Similar M. bovis diversity has been reported in Mexican dairy cattle, where even isolates from the same herd show different spoligotypes [26] and other genetic markers [27]. The presence of a spoligotype with total similarity to that of the BCG vaccine in an HIV+ patient raises important questions about the origin of the infection in particular. It was not possible to find any epidemiological connection that might explain contact with the BCG strain in this patient and currently there is no information about the presence of such a strain in Mexican cattle, in contrast with findings in French cattle infected with M. bovis, which do exhibit a BCG-like profile [28]

As previously stated, this study was performed at the Pneumonology Service of the HGM, where only adult patients are treated; therefore, it was not possible to evaluate the pediatric population, in which a high incidence of M. bovis disease has been reported [29–31]. The impact of bovine TB in the Mexican human population has been partially evaluated [27, 32, 33]; however, M. bovis TB cases in individuals born in Mexico and now living in Mexico–United States border communities [29–31, 34], and in New York City [35], have been documented.

The mode of transmission of M. bovis in the patients was difficult to determine with certainty. Although reactivation of infection acquired in childhood and person-to-person airborne transmission were possible causes, the presence of exclusively EP manifestations without clinical or radiographic evidence of pulmonary involvement, the lack of epidemiological links among the cases, and the absence of identical spoligotypes in the isolates, pointed to the ingestion of contaminated milk or dairy products as the most probable origin of the infection, as has been generally accepted [8, 36]. Socio-economic profiles of the all the M. bovis-infected patients identified in this study corresponded to those of persons of low socio-economic status living in conditions of poor housing and sanitation in the mixed rural–slum transitional areas surrounding the metropolitan area of Mexico City. Although pasteurized milk and milk products, as well as controlled cattle meat, are generally available all over the country as a result of the strict application of government-supported specific programs and control measures [10], bovine tuberculosis still persists [33, 37], and it has been reported that up to 30–40% of the dairy cattle in Mexico are infected by M. bovis [26, 33]. In the backyard stables that remain clandestinely in Mexico City, cheaper unpasteurized raw milk or dairy products are freely sold directly to consumers on the doorstep or in local street markets. A strong cultural tradition among persons with a low educational level is the preference to buy the milk direct from the stable, because they enjoy the advantage of seeing that the product comes from the farm and the belief that such milk is healthier and has a better taste [38].

The persistence of M. bovis transmission and the association with diverse causes of immunosuppression related to the living conditions of the individuals at risk, along with malnutrition, diabetes mellitus, and concomitant infections, mainly HIV, undoubtedly increase the possibility of disease progression after infection.

In conclusion, our results revealed that M. bovis persists as an important etiological agent in Mexico, despite the efforts of the specific eradication program, and should be considered a persistent agent. This agent should be specifically investigated in the differential diagnosis of EP disease in every suspect patient, particularly in those with a history of consuming unpasteurized dairy products, because the opportune identification of M. bovis will be an invaluable tool to ensure that the appropriate therapeutic scheme is applied. The strict application of surveillance measures in cattle, milk, and milk products, as well as intensive educational campaigns against the consumption of products from unpasteurized cow’s milk should be encouraged.