Background

Mycobacterial diseases are a major health concern, with an estimate of approximately one-third of the world's population being affected by them [1, 2]. Pulmonary tuberculosis, the most common clinical form of the disease, is a granulomatous disease of the lungs caused by Mycobaterium tuberculosis. However, only 5-10% of the infected people ever develop the disease. The genetic contribution of the host plays a significant role in determining susceptibility to developing the active form of the disease, the severity of infection and the health outcome of the patient [3, 4]. A number of genes have been identified in studies of diverse origins to be important in tuberculosis [5–7].

The genes for tumor necrosis factor-α (TNF-α; TNFA) and lymphotoxin-α (LT-α; LTA), located within the MHC III region of chromosome 6, shows close linkage to the HLA class I (HLA-B) and class II (HLA-DR) genes [8]. Both TNF-α, produced mainly by monocytes and activated macrophages; and LT-α, produced mainly by activated T-cells, play important immunoregulatory roles [9]. Of these, TNF-α contributes to the pathogenesis of tuberculosis due to its role in the formation and maintenance of granulomas [10]. Additionally, it also plays a major role in host defense to M. tuberculosis by its synergistic action with interferon-γ (IFN-γ) to activate macrophages and thereby impacts on disease perpetuation [11, 12]. Elevated serum TNF-α (sTNF-α) levels have been reported in advanced tuberculosis patients when compared to those with mild tuberculosis and healthy individuals [13].

Studies on monozygotic twins and their first-degree relatives, using ex vivo endotoxin stimulated whole blood samples, have provided evidence that 60% of variation in the production capacity of TNF-α appears to be genetically determined [14]. Several polymorphisms within the promoter region of TNFA and the intron 1 polymorphism of LTA, in particular, have been associated with altered levels of circulating TNF-α [15, 16]. A few of these polymorphisms have been also studied for determining susceptibility or resistance towards tuberculosis in several ethnic groups, the results of which have been inconclusive [17–26].

The aim of this study was to determine associations, if any, of potentially functional TNFA and LTA polymorphism(s), both individually and at the haplotype level, with tuberculosis in patients from North-India. In addition, we have attempted to explore whether any of these polymorphisms may be related to the severity and the associated features of the disease. We also attempt to correlate these polymorphisms with sTNF-α levels in the patients and the controls.

Methods

Study Subjects

185 unrelated tuberculosis patients (mean age 32.16 ± 13.8 years; male:female 0.42:0.58), who presented to the Medical Outpatient Department or were admitted at the All India Institute of Medical Sciences hospital, New Delhi, between 1996 and 2006 were recruited. The study was approved by the ethical committee of the hospital and informed consent was obtained from all participants. Patients with chronic illness such as cirrhosis of liver, chronic hepatitis, acute viral hepatitis, and/or gastrointestinal, renal, or cardiac diseases and those who were tested positive for HIV using ELISA were excluded during the preliminary evaluations.

Pulmonary tuberculosis

There were 149 patients in this group (Table 1). The diagnosis of pulmonary TB was based on the presence of acid-fast bacilli on sputum smear or M. tuberculosis on sputum culture. In patients with smear-positive TB, culture was not done unless multidrug- resistant TB was strongly suspected. Sputum cultures were done in all patients with smear-negative pulmonary TB. In patients with negative smears and cultures (N = 17), the diagnosis of TB was based on symptoms, chest radiographic infiltrates in the upper lobes, and clinical and radiographic response to antituberculosis drugs. In addition, high resolution computed tomography (HRCT) for the patients showed tree-in-bud appearance compatible with the clinical diagnosis of pulmonary TB. Also, all 149 pulmonary TB patients showed a good clinical and radiographic response to anti-TB treatment. The severity of the pulmonary lesions was graded by studying the chest radiographs following the guidelines of the American Thoracic Society [27].

Table 1 Clinical characteristics of the tuberculosis patients used for the study
Full size table

Disseminated tuberculosis

Focal disseminated tuberculosis (DTB) was diagnosed based on the criteria including i) clinical features suggestive of tuberculosis; ii) concurrent involvement of at least two non-contiguous organ sites of the body or involvement of the bone marrow; iii) microbiological and/or histopathological evidence of tuberculosis; (iv) evidence of marked improvement on antituberculosis treatment. The presence of all criteria was required for the diagnosis of disseminated tuberculosis and there were 11 patients in this group (Table 1). Of these, 2 patients had positive Ziehl-Neelsen (ZN) staining on fine needle aspiration cytology (FNAC) of the lymph nodes; 3 patients had histopathology of lymph nodes compatible with the diagnosis of TB; 4 patients had both positive ZN staining and Mycobacterium tuberculosis (Mtb) culture positivity in the pus from cold abscess and additional 2 patients had only Mtb culture positivity. All patients showed a good clinical response to anti-TB treatment.

Lymph node tuberculosis

Lymph node tuberculosis (LNTB) was diagnosed based on the following criteria i) clinical presentation compatible with tuberculosis; ii) fine needle aspiration cytology presenting evidence of tuberculosis with positive staining for acid-fast bacilli; iii) histopathological evidence of caseating granulomas on lymph node biopsy specimens; iv) evidence of marked improvement on antituberculosis treatment. The presence of i), and either ii) or iii) and iv) was required for the diagnosis of lymph node tuberculosis and there were 25 patients in this group (Table 1). Of these, 7 patients had positive ZN staining on FNAC and 20 had histopathology of the lymph nodes compatible with the diagnosis of tuberculosis of the lymph nodes (2 of these also had positive FNAC). All these patients showed a good clinical response to anti-TB treatment.

Normal Controls

Healthy Volunteers (referred to as normal controls, N = 155; mean age 28.16 ± 13.6 years; male:female 0.46:0.54) were randomly recruited from the general population with the same socio-economic status and ethnic background as that of the patients. All individuals gave informed consent and were screened negative for HIV infection and a family history of tuberculosis or any other related disease. In addition, these individuals had a negative tuberculin test (to exclude a possibility of a latent tuberculosis infection) and their chest radiographs as well as peripheral blood counts and blood chemistries were normal.

All patients with TB and normal controls recruited for the study represented a fairly homogeneous ethnic group of north Indian Hindu population from the states of Punjab, Haryana, Uttar Pradesh and Delhi. All these individuals come under the Indo-European linguistic group and belong to the same ethnic group (group 1) [28, 29].

Tuberculin Skin Test (TST)

The TST was done using 5 tuberculin purified protein derivative (PPD) units (5 TU Tuberculin PPD/0.1 ml; Span Diagnostics Ltd., India); and the indurations were measured after 48 hours. A cut-off of 10 mm was used for a positive TST.

Serum TNF-α measurement

Serum TNF-α levels were determined for 107 tuberculosis patients and 112 normal controls using the OPTEIA™ ELISA kit (BD Biosciences), as per the manufacturer's instruction.

Genotyping

Isolation of the genomic DNA from the peripheral blood leukocytes was carried out using the modified salting out procedure as described elsewhere [16]. Polymorphisms in the TNF genes, namely TNFA-1031T > C, TNFA-863C > A, TNFA-857T > C, TNFA-308G > A, TNFA-238G > A and the LTA_NcoI polymorphism were genotyped using primer pairs as described elsewhere [16]. Briefly, the TNFA-1031T > C, TNFA-863C > A, TNFA-857T > C, TNFA-308G > A and the TNFA-238G > A were studied using SNaPshot ddNTP Primer Extension Kit (Applied Biosystems, Foster City, USA) as per the manufacturer's instructions. To clean up the primer extension reaction, 1U of calf intestinal phosphatase (CIP) (New England Biolabs) was added to the reaction mixture and the mixture was incubated at 37°C for 1 hour, followed by 15 minutes at 72°C for enzyme inactivation. These samples were subsequently electrophoresed using the ABI Prism 3100 Genetic Analyzer as per the manufacturer's instructions. The results were analyzed using the GeneMapper software (Applied Biosystems, Foster City, USA). The LTA_NcoI polymorphism was assessed using NcoI restriction endonuclease digestion. The accuracy of genotyping was confirmed by direct sequencing of the DNA samples (N = 5) for all three respective genotypes for all the loci investigated.

Statistical Analysis

The allele frequencies were calculated and agreement with Hardy-Weinberg equilibrium was tested using a χ2 goodness-of-fit test for each locus. The association between two categorical variables was evaluated by likelihood ratio (LR) χ2 and Fishers exact tests. All statistical tests performed were two-tailed. ANOVA was performed to test the effect of the polymorphisms and haplotypes on sTNF-α levels and F-ratio statistics were calculated to study the variation in sTNF-α levels in different groups. PHASE was used to infer haplotypes for each individual [30]. Clump22 software was used with 1,000,000 Monte-Carlo simulations to determine the primary difference in the haplotype frequencies [31]. Normal (T1) χ2 values, calculated from the 2 × N raw data, and maximum (T4) χ2 values, calculated by simulating 2 × N data followed by clumping into 2-by-2 tables to produce maximal chi-squared values, have been reported.

Results

Association of TNFpolymorphisms with tuberculosis

The genotype frequencies for all the polymorphisms investigated did not deviate significantly from Hardy-Weinberg expectations in both the patient and the control populations (p > 0.05). Table 2 and Table 3 enlist the allele and the genotype frequencies for all the polymorphisms genotyped in the TNF genes, respectively. The comparison of the allele and genotype frequencies for all the loci investigated revealed no significant differences between the tuberculosis patients and controls (p > 0.05; Table 2 and Table 3, respectively).

Table 2 Allele frequencies of the polymorphisms investigated in the TNFA and LTAgenes.
Full size table
Table 3 Genotype frequencies of the polymorphisms investigated in the TNFA and LTAgenes.
Full size table

Association of TNFpolymorphisms in tuberculosis sub-groups

The severity of the pulmonary lesions was graded by studying the chest radiographs following the guidelines of the American Thoracic Society [27]; the patients were sub-grouped into minimal, moderately advanced and far advanced. Further, the patients were also sub-grouped on the basis of response to TST (negative/positive) and the presence/absence of cavitary lesions. However, none of the studied polymorphisms showed a significant association with any of the patient sub-groups (p > 0.05; data not shown).

Association of TNFhaplotypes with tuberculosis

To study the combined effects of the six SNPs in the TNF genes, haplotypes were estimated by the statistical software PHASE. Of the 13 haplotypes observed in the 185 TB patients and 155 controls recruited for the study, only five haplotypes were represented at frequencies >5% (Figure 1). These accounted for >87% of the total haplotypes. As the count per haplotype was very low, Clump22 with 1,000,000 Monte-Carlo simulations was used to estimate differences in the haplotype frequencies in the patients and controls. However, no significant differences were observed in the haplotype counts in the two groups [normal (T1) χ2 = 18.498, df = 12, P = 0.78; and maximum (T4) χ2 = 7.825, P = 0.214]. Therefore, it may be concluded that haplotypes in the TNF genes were not associated with tuberculosis in the studied population.

Figure 1
figure 1

The frequency distribution of the haplotypes of the TNFA and LTA genes. The frequency distribution of the haplotypes of the TNFA and LTA genes in tuberculosis patients (dark bars) and unrelated controls (light bars). The haplotypes have been plotted on the x-axis and their respective relative frequencies (%) on the y-axis.

Full size image

Association analyses of TNFpolymorphisms with its serum levels

TNF-α levels were measured in the serum samples of 107 tuberculosis patients and 112 unrelated controls. A difference with borderline significance was observed in the serum TNF-α levels in the patients and the controls (F-ratio = 3.84, df = 1, p = 0.051); a higher levels of the cytokine was observed in the patients sera [log sTNF-α ± standard deviation (s.d.) = 1.89 ± 0.87] when compared with the control population [log sTNF-α ± s.d. = 1.65 ± 0.89].

Although, none of the investigated polymorphisms were found to affect the sTNF-α levels in both the patient and the control populations and in the patient sub-groups (p > 0.05), a few interesting observations were made. Of the 107 patients, 39 patients with positive TST showed higher log sTNF-α levels (2.10 ± 0.91) while 68 patients with negative TST showed lower log sTNF-α levels (1.75 ± 0.84) [F-ratio = 3.92, df = 1, p = 0.05]. Furthermore, when the patients were graded on the basis of severity into minimal severity, moderately advanced and far advanced sub-groups, it was observed that patients with minimal severity had lower log sTNF-α levels (1.54 ± 0.73) when compared to the patients with moderately advanced (1.86 ± 0.98) and far advanced severity (1.96 ± 0.71). However, none of these differences were statistically significant (p > 0.05). Additionally, no difference was observed in the log sTNF-α levels when patients were sub-grouped on the basis of presence/absence of cavitary lesions in the X-rays (p > 0.05).

Furthermore, when haplotypes (counts ≥ 2) were analyzed with respect to sTNF-α levels in the patient samples (case-only analysis), no significant difference were observed (p-value > 0.05), thereby negating the influence of any of the studied polymorphisms on sTNF-α levels in the study population.

Discussion

TNF-α plays an important role in granuloma formation in tuberculosis. Previous studies have shown that mice deficient in TNF-α exhibit poorly formed granulomas with areas of extensive necrosis, resulting in widespread dissemination of M. tuberculosis and rapid death of animals [32]. Earlier data obtained from the rheumatoid arthritis patients treated with TNF-α antagonists showed that blocking TNF-α could lead to reactivation of tuberculosis [33]. All these points hint towards TNFA as an important candidate gene for tuberculosis.

The promoter region of the TNFA gene is highly polymorphic. In this case-control study, conducted in tuberculosis patients and normal controls recruited from Northern India, we focused on potential functional polymorphisms in the transcriptional region of TNFA and a LTA_NcoI (+252) polymorphism in the first intron of LTA gene a few of which have earlier been associated with TNF-α levels [15, 16]. Individually, none of the polymorphisms showed any significant association with tuberculosis or sTNF-α levels. An earlier study carried out in HLA-A, B and DR typed pulmonary tuberculosis patients (N = 210) and healthy control subjects (N = 120) from South India also failed to detect any significant differences between the LTA_NcoI and TNFA_-308 and TNFA_-238 polymorphisms and tuberculosis [20]. These included active and cured tuberculosis patients as well as controls of Indo-dravidian origin, an ethnicity different from ours, recruited from North-India and belonging to the Indo-European linguistic group. Additionally, another recent study carried out in the north-Indian populations also failed to find any association of TNFA -308 promoter polymorphism with tuberculosis [34]. This study was conducted in ethnically matched 145 physician diagnosed tuberculosis patients and 211 normal healthy controls recruited from the state of Punjab from North India. However, in both these studies no attempts were made to correlate these and the other promoter polymorphisms individually and at the level of haplotype with sTNF-α levels. Also, the normal controls recruited for our study were tested negative by the TST test and had normal blood chemistries, peripheral blood counts and normal chest radiographs. Additionally, all patients and controls recruited for our study were HIV-negative. Our study also gains support from others which report a lack of association between TNF gene polymorphisms and tuberculosis in different populations (Table 4). On the other hand, a few studies reported association between the TNFA_-308 polymorphism and tuberculosis [22, 35–37]. However, these differences in results could be attributed to ethnic variations that persist in different populations, study design and patient selection criteria. Additionally, the association observed in the later studies may be explained on the basis of linkage disequilibrium that exists between the TNF genes and HLA [38].

Table 4 Association studies of the TNF genes with tuberculosis.
Full size table

Our study also failed to detect any effect of the TNF polymorphisms on sTNF-α levels in the tuberculosis patients. The results from previous in vivo and in vitro studies conducted for the TNFA-308G > A polymorphism have rather been conflicting. Whereas there are studies which report the minor TNFA-308A allele (allele 2) to be associated with higher inducible levels of gene transcription and TNF-α production [39, 40], our earlier study in sarcoidosis patients found the G allele for this polymorphism to be associated with high sTNF-α levels [16]. It is also evident from the study by Knight et al that the TNFA-308 SNP does not regulate TNF-α levels and is not likely to be the functionally important SNP as previously hypothesized [41]. Thus, it may be possible that another SNP in LD is involved in this phenomenon. Further high resolution HLA typing data in addition to fine mapping of the TNF genes would be of help in resolving these problems.

To enhance our understanding about the contribution of these genetic variants further to tuberculosis, six-locus haplotypes were constructed and their distribution was compared in the patients and the control population. None of the haplotype showed significant association with tuberculosis in the study population.

Conclusions

In summary, our findings suggest that the polymorphisms present in and around the TNF genes are unlikely to be the major risk factors for tuberculosis in Asian Indians. However, as the present study was performed with relatively small sample size, further studies with larger sample sizes would be necessary to elucidate the role of TNF polymorphisms in tuberculosis.