Our case-control study provides evidence for a robust association between high HAP exposure from biomass smoke and risk of TB among adult PLHIV. Using a validated questionnaire, we found that high daily and cumulative exposure to HAP increased TB risk. Based on domestic CO measurements, we showed an exposure-response relationship between CO concentration and PTB risk, with participants in the highest quintile of 24h personal CO TWA concentrations exhibiting a fourfold higher risk of having developed TB than those in the lowest quintile, independently of other variables.
We found that “time spent in the kitchen” was a predictor of PTB, particularly for women, despite the well-known, protective role of female gender against TB [30–32]. When stratified by “years since first ugali”, the variable “time spent in the kitchen” remained independently associated with PTB among participants in the highest stratum (> 25 years cooking ugali). For younger participants with < 25 years since first ugali, using only the traditional “three stones” for kitchen activities increased PTB risk. However, among these younger cooks, cooking inside the main house (a proxy for using a non-biomass energy source) was associated with a reduced PTB risk, especially if a chimney was present. In Africa, people in urban or suburban areas typically have two kitchens – an indoor one using electricity or gas, and an outdoor one using solid fuel. However, in the DRC, ownership of electrical appliances (56% in our study) is not a good indicator for low HAP exposure, because an inconsistent power supply forces many to rely for much of the time on non-electricity sources, such as biofuel[33]
Our findings from the questionnaire were largely corroborated by personal CO measurements, which were obtained for 20% of participants. In this subgroup, TB risk was significantly associated with “time spent in the kitchen”. In addition, there was a significant and dose-dependent association with TWA 24h personal CO, but not with maximum 24h personal CO levels. The lack of association between maximum CO and PTB suggests that prolonged exposure is needed to affect the risk of PTB.
HAP and tobacco smoke similarly affect human health. Meta-analyses have shown that tobacco smokers are about twice as likely to develop (fatal) PTB than non-smokers.[34–36] Over the last decade, conflicting data have been published about the relationship between HAP exposure from biomass and PTB risk.[17, 37] Two systematic reviews found very low-quality evidence for an additional risk of TB in relation to HAP exposure. [38, 39] Conversely, an association between solid fuel use and TB was found in an analysis of thirteen studies conducted between 1996 and 2012.[40] Ten of these studies yielded a pooled OR of 1·30 (95% CI 1·04–1·62), while six yielded a pooled OR of 1·70 (1·10–8·20) in a subgroup analysis considering gender. A meta-analysis[41] including twelve studies reported an overall effect estimate of 1·43 (95% CI 1.07–1·91) and, among women, 1·61 (95% CI 0·73–3·57). Finally, a 2020 global systematic review (53 studies) estimated a pooled relative risk of 1·26 (95%CI 1·08 − 1·48) of PTB associated with HAP exposure6. This association has been less studied among PLHIV, who are the most vulnerable to TB.[2, 6, 17] The lungs have been described as an anatomic reservoir of HIV,[42] and tobacco smoking is known to induce pulmonary immune defects.[43] Interactions between TB-HIV-tobacco smoking and non-communicable lung diseases have been demonstrated in several studies and summarized in two reviews by van Zyl-Smit and colleagues.[10, 11]
The pathophysiologic mechanisms causing lung illnesses after HAP exposure are not fully understood and while combustion might produce both CO and PM2.5, their ability to impair cell immunity might be different. The current evidence pertaining to HAP mechanistic effects was summarized by the European Respiratory Society/American Thoracic Society task force on HAP.[8] Based on this expert panel that reviewed both cell culture and animal studies, it appeared that numerous HAP-related health effects are also a consequence of impaired bacterial phagocytosis in alveolar macrophages loaded with carbon.[43, 44] From ambient air pollution studies, we know that human -defensin 2 and 3 expression in M. tuberculosis-infected A549 cells was reduced by exposure to PM2.5 or PM10[45]. The ability of cells to control M.tb growth and the M.tb-induced expression of CD69, an early surface activation marker expressed on CD3 + T cells, as well as the production of IFN-, TNF-, and TBX21 in M.tb-infected PBMC, were all reduced when exposed to PM2.5 prior to M.tb infection[46]. This suggests biological pathways underpinning changed M.tb infection and treatment results when exposed to PM2.5[47].
Time spent in the kitchen proved to be a determinant of risk for TB. In addition to doubling the risk of developing TB among PLHIV, long-term tobacco smoking has been shown to attenuate both immune and antiviral responses to antiretrovirals by as much as 40%.[9] At the cellular level, our findings could partly be explained by recent evidence that cytokine production by alveolar macrophages (AM) is inversely related to chronic biomass smoke exposure.[8] Moreover, the association we observed between CO exposure and PTB might be explained by impaired oxidative responses.[8, 44, 48] Compared to PM, it is know that lung and systemic M. tuberculosis-induced cytokine production are altered by PM load in AM and that chronic PM exposure with elevated proinflammatory cytokine expression leads to cellular inactivity.[49] In addition, the pulmonary compartment contains many macrophage-specific immunological deficiencies in smokers, which may explain how smoking makes a patient prone to TB infection and illness.[50] As a result, cigarette smoke reduces effector cytokine responses and inhibits mycobacterial containment inside infected human macrophages from the peripheral blood and alveolar compartments.[51] After Mtb infection, human AM show metabolic plasticity that facilitates glycolytic reprogramming. Smokers also have reduced metabolic reserve, impairing the glycolytic response to infection.[52] Our results, showing a 1·4-fold increase in the odds of developing PTB after chronic high exposure to biomass smoke, are compatible with these findings. It is also wellestablished that level of impairment is more severe with exposure to wood smoke than to fine carbon black, [8] and we demonstrate here that exclusive wood use for cooking increased the likelihood of having developed TB approximately fourfold among younger cooks with HIV infection. Future mechanistic studies (causal mechanisms between the environment and the host response to tuberculosis) should consider HIV-infection status (level of immune defence vs pollutant dose-response) and evaluate if other pollutants (multipollutant model), such as volatile organic compounds (e.g., benzene metabolites), also interfere with mechanisms affecting PTB (risk, new Mycobacterium tuberculosis rate, clinical outcomes) as recently suggested by a study on latent TB infection in Vietnam where PM2.5 did not show a significant association as expected.[53, 54]
During the past decade, major randomized trials in LMICs, such as the RESPIRE (Guatemala)[55], CAPS (Malawi)[56], and currently the GRAPHS (Ghana)[57] trials, have used CO as a surrogate for HAP exposure. Similar to our findings, a large nested case-control, single pollutant study in California (2,309 cases and 4,604 controls)[19] found an association between quintiles of CO and TB risk, whereas no association was found for quintiles of PM2.5. However, it is known that CO and particulates do not always correlate.[58] We need simple, affordable, and reliable markers of exposure 4,8,34 to inform well-designed interventional studies to reduce HAP-induced chronic lung disease.
As part of worldwide TB control efforts, initiatives to integrate tobacco cessation with air pollution reduction (e.g., supply of inexpensive clean energy sources, identification, monitoring, and reduction of air pollution sources) should be addressed. Such initiatives might involve developing patient- and community-focused air pollution mitigation methods and interventions in collaboration with governmental entities, such as patient-screening tools for air pollution risk in at high-risk patients for TB (eg., PLHIV by targeting woman and patients deeply immunocompromised). Interventions should address technology (e.g., improved solid fuel stoves such as traditional and modern combustion designs using fans or gasification equipment), fuel type (e.g., unprocessed or processed such as pellets for biomass, briquettes for coal as well as cleaner fuels such as liquid petroleum gas, biogas, permanent electricity, solar lights), better ventilation system (e.g., chimneys, opening windows while cooking), and behavioral adjustments (e.g., enables for cleaner technologies and fuels such dryness of fuel and community endorsement by community leaders or religious).[17, 60] In light of the current COVID-19 pandemic that has hampered several efforts to eradicate TB, preventing future TB requires addressing not just the disease but also the major drivers of TB (undernutrition, poverty, diabetes, cigarette use, and household air pollution) if TB is to be eradicated by 2035.[61] Hence, our study has significant implications for addressing the global respiratory health threat in LMICs, which is fuelled by a high prevalence of chronic respiratory diseases (asthma, chronic obstructive pulmonary disease, bronchiectasis, and post-tuberculosis lung disease), COVID-19 infection and long COVID-19, all of which are associated with environmental factors and endemic HIV.[62–65] In Box 1, we have summarized and contextualized the implications of our findings in the fight against tuberculosis. Several further strengths of our study include the following: : 1) A large sample size, which adequately powered for our primary objective to investigate HAP-associated PTB risk in PLHIV; 2) Combination of a validated IMPALA questionnaire and 24h personal CO monitoring data; 3) Quantification of cumulative exposure to HAP using simple indexes adaptable/generalizable to other LMIC settings; 4) Documentation of an exposuredependent relationship between HAP and TB risk; 5) Findings suggesting increased PTB risk for women, as they are more likely to be exposed to HAP in our DRC setting.
Although informative, our study also has several limitations. Questionnaire responses and our proxy for lifetime exposure may have been affected by recall bias; however, it is unlikely that responses were influenced by case or control status, both of which were defined using reliable objective criteria. Besides, such recall bias would tend to dilute the effect size, thus making our findings conservative and hence more compelling. Nevertheless, to mitigate recall bias, trained interviewers used the IMPALA questionnaire, and multiple sources were used to triangulate information. We did not account for unmeasured confounding, such as nutritional status and overcrowding, but included surrogates for socio-economic status, such as educational attainment and income. We also relied on one-time 24h personal CO measurements, which might underestimate or overestimate true effect size. However, recent data have linked short term (three-month) exposure to air pollutants (PM, CO, etc.) and increases in TB incidence.[20] We did not consider meteorological factors, but these are unlikely to have introduced systematic biases as the weather varied little during the study, and CO measurements were consistently taken during the same time-period for cases and controls. As indicated before, we acknowledge that CO may not be the best indicator of HAP exposure from biomass smoke. It would have been desirable to measure fine particulate matter, possibly the most harmful component of biomass smoke,[29] or other biomarkers of wood smoke exposure such as urinary guaiacol or levoglucosan.[8] In the absence of ambient air quality monitoring in South-Kivu – as in most areas in Africa[66] – we did not take outdoor air pollution into account. However, ambient air pollution is unlikely to have differed between cases and controls because both were recruited from the same small geographical area. Future research should explore long-term exposure monitoring and/or a biomarker of HAP exposure. However, a multipollutant model that measures both PM2.5 and CO utilizing a low-cost, long-lasting battery sensor may be beneficial for orienting intervention in environments with variable resources. Finally, there was unequal gender distribution,[22] due to the high proportion of women among PLHIV, on one hand, and higher risk of having TB among men, on the other. We addressed this issue by performing sex-stratified analyses.
In conclusion, personal CO exposure and time spent cooking (among women) were found to be independently associated with increased odds of PTB among PLHIV. Public health implications of our findings, if confirmed by further longitudinal studies, are that HAP interventions might prove cost-effective if reductions in TB are considered and measured since HAP is modifiable. Furthermore, progress in the fight against TB might be stalled if we do not adequately address HAP, which is an increasing problem in LMICs with high burdens of both HIV and TB.