Genetic variability and consequence of Mycobacterium tuberculosis lineage 3 in Kampala-Uganda

Background Limited data existed exclusively describing Mycobacterium tuberculosis lineage 3 (MTB-L3), sub-lineages, and clinical manifestations in Kampala, Uganda. This study sought to elucidate the circulating MTB-L3 sub-lineages and their corresponding clinical phenotypes. Method A total of 141 M. tuberculosis isolates were identified as M. tuberculosis lineage 3 using Single nucleotide polymorphism (SNP) marker analysis method. To ascertain the sub-lineages/sub-strains within the M. tuberculosis lineage 3, the direct repeat (DR) loci for all the isolates was examined for sub-lineage specific signatures as described in the SITVIT2 database. The infecting sub-strains were matched with patients’ clinical and demographic characteristics to identify any possible association. Result The data showed 3 sub-lineages circulating with CAS 1 Delhi accounting for 55% (77/141), followed by CAS 1-Kili 16% (22/141) and CAS 2/CAS 8% (12/141). Remaining isolates 21% (30/141) were unclassifiable. To explore whether the sub-lineages differ in their ability to cause increased severe disease, we used extent of lung involvement as a proxy for severe disease. Multivariable analysis showed no association between M. tuberculosis lineage 3 sub-lineages with severe disease. The risk factors associated with severe disease include having a positive smear (OR = 9.384; CI 95% = 2.603–33.835), HIV (OR = 0.316; CI 95% = 0.114–0.876), lymphadenitis (OR = 0. 171; CI 95% = 0.034–0.856) and a BCG scar (OR = 0.295; CI 95% = 0.102–0.854). Conclusion In Kampala, Uganda, there are three sub-lineages of M. tuberculosis lineage 3 that cause disease of comparable severity with CAS-Dehli as the most prevalent. Having HIV, lymphadenitis, a BCG scar and a smear negative status is associated with reduced severe disease.

. The M. tuberculosis lineage 3 (MTB-L3), also known as the Central Asian strains (CAS), occurs predominantly in areas around the Indian Ocean, East Africa and India [4,5]. The genetic diversity of the CAS can be defined based on specific single nucleotide polymorphisms (SNPs) [6,7], genomic deletion, also known as long sequence polymorphism (LSP) [4,5], and a particular spoligotype pattern [8]. The latter can further subdivide the main M. tuberculosis lineage 3 into specific sub-lineages [8]. Emergence and spread of M. tuberculosis lineages to other niches (where they were originally absent) has been associated with immigration, clinical and demographic factors, as well as evolution of MTB strains [9,10]. Understanding mechanisms shaping transmission of MTB strains can provide a lead about the potential approaches for TB control.
The data from our previous studies showed that in Kampala, Uganda, there are 3 main M. tuberculosis lineages circulating, of these 11% were M. tuberculosis lineage 3 [11]. Moreover, findings also revealed that all the M. tuberculosis predominant in Kampala were equally virulent (based on cavitation as a proxy for virulence). Nevertheless, elsewhere authors have reported that different M. tuberculosis complex lineages infections present with specific clinical phenotypes [3]. The failure to demonstrate specific clinical outcomes in our earlier dataset might be attributable to comparing genetically heterogeneous M. tuberculosis complex main lineages; this could have confounded our results thereby suggesting no difference in virulence. Differences in bacterial characteristics have provided insight into how the M. tuberculosis complex bacteria cause disease, and why some are geographically wide spread. For instance, the Beijing strains that belong to M. tuberculosis lineage 2 are highly virulent, prone to drug resistance and BCG vaccination is not protective. This may partly explain why they are a global threat [12][13][14][15]. Additionally, strains of M. tuberculosis lineage 4 are associated with pulmonary tuberculosis and severe lung consolidation, less virulent [16] and prone to anti-tuberculosis drug resistance [17] as opposed to other sub lineages. Similarly Newton et al, [18] showed that sub-lineages of M. tuberculosis lineage 3 cause severe disease; Stucki et al, [19] and Hershberg, 2016 [20] showed that M. tuberculosis lineage 5-7 have a narrow host range, thus they are restricted to particular geographical niche. Therefore, accurate understanding of M. tuberculosis complex sub-lineages and their clinical outcomes can bolster the development of appropriate intervention strategies that more effectively target the circulating strains. Given that background in the current study, we are describing sub-lineages/sub-strains within the main M. tuberculosis lineage 3, the least dominant MTB lineage in kampala. To answer this question we shall start by analyzing the MTB direct repeat (DR) loci for sub lineages within M. tuberculosis lineage 3 as well as understanding the demographic and clinical manifestation of patients infected with MTB-L3 sub lineages. With such an approach, we can describe whether sub-lineages of M. tuberculosis lineage 3 prevalent in Kampala, Uganda differ in their ability to cause severe disease (extent of lung involvement abnormalities) as evaluated by chest x-ray.

Study design and M. tuberculosis isolates
The M. tuberculosis isolates used in this study were obtained from adult (� 18 years) patients (index cases) and their household contacts (HHCs), confirmed with pulmonary TB by culture in a cross sectional study (2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012) in Kawempe division Kampala, Uganda [11,21], where the data for the current study is coming from. The HHCs were TB patients who had stayed with an index patient for at least 7 consecutive days for the previous 3 months. The index cases residing with 1 or more HHCs were enrolled in the study through the clinic at the Uganda National TB and leprosy program at Mulago Hospital or by referral to the TB research clinic at Mulago Hospital or through public sensitization in Kawempe division. Adults with clinical signs (a positive chest x-ray or sputum smear positive) suggestive of tuberculosis provided a sputum sample for culture following standard laboratory procedures. The patients with active TB were treated using a short course therapy of Isoniazid (INH), rifampicin (RIF), pyrazinamide and ethambutol for 2 months, followed by 4 months of INH and RIF. The cultured samples were later tested for drug resistance, patients with resistant MTB isolates were provided with treatment according to the TB program guidelines. The HHCs � 5 years old, HIV and TST-positive were prophylactically treated with INH for 6-9 months. Patients' baseline demographic and clinical variables such age, sex, HIV status, employment status, status on income, TB cavitation on chest x-ray (present or absent), ethnicity (Bantu & others), status of smoking, body mass index (BMI) calculated from height & weight, alcohol drinking, presence of BCG scar, whether patients have night sweats, knowledge of TB in the past, presenting with hemoptysis (cough with blood), having swollen lymph nodes (lymphadenitis), evaluation of extent of lung involvement on chest radiography (classified as normal, mild, moderate, or far advanced) and smear status (positive or negative), were recorded by a medical physician or a laboratory technician.

Genomic DNA extraction and genotyping M. tuberculosis isolates
DNA extraction for 141 M. tuberculosis isolates and SNP (lineage-specific SNP for M. tuberculosis lineage 3: Rv0129c_0472n) typing to identify M. tuberculosis lineage 3 was performed as described by Wampande et al, [11]. To determine the sub-lineages of M. tuberculosis lineages 3, the isolates were further analyzed with a spoligotyping commercial kit as described by Kamerbeek et al, [22], the shared international type (SIT) spoligotyping were assigned according to SITVIT and SITVIT2 database [8,23]. lungs with no cavitation) or advanced disease (lesions more extensive than minimal disease with cavitation) on chest x-ray examination [24]. Univariate analysis was perfomed and the chi square test or Fisher's exact test was used to compare the distribution of categorical variable by disease. Variables in univariate analyis with P � 0. 2; except HIV a known risk factor for TB, were included in the multivariable logistic model. Multivariable logistic regression was used to evaluate the association between sub-lineages (sub strains) of M. tuberculosis lineage 3 (independent variable) and extent of lung involvement (minimal or advanced) disease on chest x-ray (dependent variable). The 2 individuals infected with CAS were excluded from the analysis because of the small number. Age, sex, smear status, HIV status, BCG scar, smoking status, swollen lymph nodes (lymphadenitis) and BMI were used as adjusters. All analyses were conducted with Stata software, version 12 (StataCorp, College Station, Texas).

Ethics
The institutional review boards and ethics committees at University Hospitals of Cleveland, Makerere University, and the National HIV/AIDS Research Committee as well as the Uganda National Council for Science and Technology approved the study protocols. All patients gave written informed consent for study participation, including pre-and post-HIV test counseling.

Demographic and clinical characteristics of the study participants
For the analysis we included 141 M. tuberculosis lineage 3 isolates, each corresponding to a tuberculosis patient.
The description of the patients demographic and clinical characteristics has been detailed in Table 1; the proportions of the patients' characteristics for the different variables among the sub-lineages of M. tuberculosis lineages 3 (Table 1) were generally similar irrespective of the MTB sub-lineage. From now onwards we have excluded the CAS strains in the analysis due to a small number (2 strains).

Risk factors associated with MTB lineage 3 infections
In all the analyses, CAS1-Dehli was used as the reference since is the most prevalent, and we set out to understand why it is dominant in comparison with other sub lineages circulating in the study area. Univariate analysis showed that disease severity (extent of lung involvement: minimal versus advanced disease) was not associated with any of the sub-lineages of M. tuberculosis lineage 3 (P� 0.05).

Multivariable analysis for association between severe lung disease and sub lineages of M. tuberculosis lineage 3
In the multivariate analysis after adjusting for sex, smear status, HIV status, BCG scar, smoking status and lymphadenitis, the data suggests that severity of TB disease is not dependent on the M. tuberculosis sub lineages (P � 0.05).

Discussion
M. tuberculosis infections are of global concern, therefore understanding the drivers of disease progress and spread is paramount. Host and environment factors have been suggested as key players among others that can bolster TB spread, there is also overwhelming evidence that   In our study, among sub-lineages of M. tuberculosis lineage 3, the most successful sub-lineage was CAS 1-Dehli that causes at least 50% of the pulmonary TB, followed by CAS 1-Kili and CAS. This current data is contrary to earlier findings by Asiimwe et al, [25] in central Uganda, who showed that CAS 1-Kili was the most prevalent sub-strain, yet Bazira et al, [26] in western Uganda observed only CAS-Dehli sub-strains. In another study that exclusively considered extra pulmonary TB showed CAS 1-Dehli as the most prevalent, the previous 2 studies compares well with the current data [27]. Despite these incongruences, we argue our data is more robust since spoligotyping was performed on isolates that were first confirmed as M. tuberculosis lineage 3 by SNP [7] typing. The approach of defining first the main MTB lineage by SNP typing reduces on the errors of misclassifying intra lineage sub strains by spoligotyping since the direct repeat loci is prone to convergent evolution [6]. The other studies described exclusively used spoligotyping technique alone to define the sub lineages, and this could result in misclassification of sub lineages due to convergent evolution, thereby impacting the data. Moreover, in addition to MTB-L3 sub lineages, they considered other MTB lineages in the same study, which can disproportionately misrepresent the status quo due to overrepresentation of other sub lineages in the study area [11,28]. Our current data demonstrated quite a number of isolates, 21% (30/141) that could not be classified in any of the known sub lineage. This finding leads one to consider that these might be unknown strains. Nevertheless, we cannot rule out the possibility of mixed (having more than one sub lineage) infections in patients as earlier reported by Dickman et al, [29] who studied isolates from the same study area. Such a scenario produces muddled finger prints which cannot be ascribed to any of the known shared international type (SIT) spoligotypes in the SITVIT2 database. Efforts are underway to fully characterize these supposedly "unknown strains" and have them undoubtedly described to the M. tuberculosis research community.
From our current data, to assess why CAS 1-Dehli is the most successful sub lineage in causing disease, we hypothesized that sub-lineages within M. tuberculosis lineage 3 differ in their ability of causing advanced severe disease; we defined severe disease as extent of lung engrossment with TB specific lesions and cavitation (minimal or advanced disease) on chest xray. Our data shows that the M. tuberculosis sub-lineages circulating in central Uganda equally cause disease in the infected patients (P � 0.05). The CAS-sub-lineage suggests an association with severe disease (aOR = 5.9; aCI = 0.36-95.76), but then again due to the small sample size the wide confidence interval does not support the finding, this calls for another bigger study to substantiate on this observation. Contrary to our findings, M. tuberculosis lineage 3 sub strain infections have been associated with different phenotypes for instance, reduced expression of TNFα and IFNγ, reduced growth rate in macrophages [18,30], causing cavitary TB, pan sensitivity to anti-TB drugs [31] and causing severe disease [18]. Noticeably, TB household population studies can be confounded by a number of factors that could have affected our downward data analysis [32]. Nonetheless, we think our analysis was robust enough since known risk factors, such as patients with a positive smear (OR = 9. 384; CI 95% = 2.603-33.835) were associated with severe disease, HIV reduces (OR = 0.316; CI 95% = 0.114-0.876) the risk of developing severe disease [33,34]. Additionally, the data showed that patients with BCG scar (OR = 0.295; CI 95% = 0.102-0.854) and swollen lymph nodes (lymphadenitis) were less likely to develop advanced severe disease. Presence of scar on the shoulders suggests that the patients were vaccinated with a BCG vaccine. The efficacy of the BCG vaccine has been found to be variable in conferring protection against M. tuberculosis infection [35,36]. For instance BCG vaccination is not protective to M. tuberculosis Beijing (MTB lineage 2) strains [12,37], but is protective of lineage 4 (H37RV, Harlem) and M. canetti strains [38]. This data therefore suggests that BCG vaccination might be protective against the development of advance severe disease in M. tuberculosis lineage 3 sub strains infections. Whether this is true between lineages, another study can elucidate on this observation. In addition, the data suggests that patients with lymphadenitis (OR = 0.171; CI 95% = 0.034-0.856) are less likely to develop severe disease. This could be for two reasons; perhaps patients had other infections that caused the lymphadenitis and not M. tuberculosis lineage 3 infections per say. Secondly, trafficking of M. tuberculosis from the primary foci (most often the lung depending on the route of infection) to the regional lymph nodes causes inflammation and subsequent localization of the bacillus in the lymphatic tissues a scenario referred to as extra pulmonary tuberculosis. Studies have demonstrated that M. tuberculosis sub lineages preferentially targets pulmonary (lungs) or extra pulmonary tissues (lymph nodes, bones, intestines, meninges among others) [39,40]. For instance, the Euro American lineage is associated with pulmonary tuberculosis [41], Beijing strains are associated with severe lung pathology [15], the East Africa India strains cause a less severe pulmonary disease [42] and CAS strains are more prevalent in extra pulmonary tuberculosis infections [27,43].

Limitations
Because MTB-L3 is not common in Uganda, our analyses of the sub lineages were limited by sample size, resulting in large confidence intervals and a potential loss of statistical power. Secondly, there was a selection bias (index patient) in recruitment of the patients which could inherently skew the findings. Thirdly, the study did not explore the possibilities of other comorbid diseases among the TB patients which could impact our results. Our approach could have been inferior to other genotyping techniques such MIRU-VNTR, whole genome sequencing in resolving sub lineages. However, the strength of this study is that we used a robust SNP typing assay to delineate MTB-main lineages 3, this improves on the accuracy of defining the sub lineages.

Conclusions
In Kampala, Uganda, there are sub lineages of M. tuberculosis lineage 3, of which CAS-Dehli is the most predominant. None of these is associated with increased risk of causing severe disease. Patients infected with M. tuberculosis lineage 3 strains who have lymphadenitis or have a BCG scar are less likely to develop severe disease; patients with a positive smear have a higher risk of developing severe disease" Supporting information S1