Abstract
Although tuberculosis (TB) remains a major killer among infectious diseases and the leading cause of death for people with HIV, drivers of immunopathology, particularly at the site of infection in the lungs remain incompletely understood. To fill this gap, we compared cytokine profiles in paired plasma and sputum samples collected from adults with pulmonary TB with and without HIV. We found that people with pulmonary TB with HIV had significantly higher markers of inflammation in both plasma and sputum than those without HIV; these differences were present despite a similar extent of radiographic involvement. We also found that the strength and direction of correlations between biomarkers in the blood and lung compartments differed by HIV status and people with HIV had more positive correlations than those without HIV. Future studies can further explore these differences in inflammation by HIV status across the blood and lung compartments and seek to establish how these profiles may be associated with long-term outcomes and lung health after completion of TB treatment.
Tuberculosis (TB) remains a major killer among infectious diseases and the leading cause of death for people with HIV. TB-associated inflammation aids in pathogen killing but can lead to unwanted lung damage. Although therapeutic strategies to modify host immune responses could decrease tissue damage, improve treatment outcomes, and reduce post-TB morbidity, drivers of immunopathology remain incompletely understood.
HIV is associated with impaired pulmonary epithelial integrity, chronic pulmonary inflammation (1), and altered circulating cytokine profiles in people with TB (2). Yet, few studies have compared cytokine profiles in the lungs of TB patients with and without HIV. This is important as pulmonary immune responses to Mycobacterium tuberculosis (Mtb) may differ from those in blood (3-6). To fill this gap, we compared cytokine profiles in paired plasma and sputum samples collected from adults with pulmonary TB with and without HIV.
Methods
We conducted a cross-sectional study in Johannesburg, South Africa during August 2021-April 2022. Adults with a first episode of Xpert MTB/RIF-positive pulmonary TB without rifampin resistance who had initiated treatment in the previous 8 weeks with at least with moderate or severe involvement on chest radiograph (CXR) were enrolled. Patients who were pregnant, incarcerated, had a diagnosis of COVID-19 within 90 days of screening, or had TB meningitis were excluded. HIV testing was performed for those with an unknown HIV status.
Participants completed a medical history questionnaire and, for those with HIV, a recent CD4 cell count was abstracted from local clinic records. Plasma and induced sputum were collected and stored at −80°C before shipment to the FIOCRUZ laboratory in Salvador, Brazil. Plasma and sputum supernatants were analyzed by bead-based multiplex assay with the Luminex™ xMAP™ INTELLIFLEX System (Milliplex MAP Human Cytokine/Chemokine Magnetic Bead Panel).
Descriptive statistics were stratified by HIV. For cytokine analyses, hierarchical clustering was used to display z-score normalized data using the ComplexHeatmap package (version 2.16.0). Log2-fold-changes were measured (gtools version 3.9.4) and median cytokine concentrations were compared with the Mann-Whitney U test. Correlation analysis was performed with the Spearman rank test, with significant correlations (p-value < 0.05) plotted with the circlize package (version 0.4.15). All analyses were performed in R 4.2.1.
The Institutional Review Board of Emory University and the Health Research Ethics Counsel in South Africa approved this study. All participants provided written informed consent.
Results
Among 30 participants enrolled, the median age was 32 (interquartile range [IQR] 27-40), 6 (20%) were female, median body-mass index was 18.9 kg/m2 (IQR 17.0-20.1), and 16 (53%) reported ever smoking (Table 1). Ten participants had HIV coinfection, of whom 5 were on ART. The median CD4 count was 197 cells/μl (IQR 31-246). There were 12 (40%) participants with moderate and 18 (60%) with severe CXR involvement. There were no significant differences in demographic or clinical variables by HIV status.
Median sputum biomarker concentrations were higher for 24 of 29 cytokines among participants with HIV than those without HIV, and were significantly higher for IL-1α, EGF, and VEGF (p-value < 0.05) (Figure 1A). Similarly, median plasma concentrations for 21 of 29 cytokines measured were higher in those with HIV, and this difference was statistically significant for IL-1α, IL-1RA, IL-6, IL-10, IP-10/CXCL10, G-CSF, and TNF-α (Figure 1A). Cytokine concentrations were at least 2-fold higher in sputum than plasma for 11 of 29 cytokines.
With hierarchical clustering, the heatmap also uncovered distinct and heterogeneous plasma and sputum profiles with clustering of participants according to relative plasma and sputum concentrations (Figure 1B). This analysis grouped patients with HIV together as having both higher sputum and plasma concentrations. Participants with HIV also tended to have positive correlations between cytokine concentrations in sputum and plasma, whereas correlations tended to be negative in those without HIV (Figure 1C). The four cytokines with the most divergent correlations by HIV status were eotaxin/CCL11, GCSF, IL-4, and VEGF.
Discussion
In this study of people with and without HIV who were newly diagnosed with pulmonary TB, those with HIV had significantly higher markers of inflammation in both plasma and sputum than those without HIV. As all participants had at least moderate involvement on CXR, these differences in inflammation were present despite a similar extent of radiographic involvement, suggesting that patients with TB-HIV coinfection may be particularly at risk for excessive lung inflammation. Furthermore, while correlations between plasma and sputum were generally poor, the strength and direction of correlations between blood and lung compartments differed by HIV status and people with HIV had more positive correlations than those without HIV.
Prior studies have also found that cytokine concentrations in lung samples differed from those in blood (4, 5). In addition, several studies have compared aspects of the host immune response by HIV status, including one study where people with TB-HIV coinfection had lower sputum concentrations of several matrix metalloproteinases and another where CD4 responses to Mtb were impaired among people with HIV (7, 8). However, these studies either did not report or did not control for the extent of radiographic involvement, which is associated with the degree of inflammation in TB.
Our finding that, among those with pulmonary TB, correlations between sputum and plasma cytokine levels were generally positive in people with HIV coinfection but negative in people without HIV may reflect impaired integrity of the lung epithelium and a propensity towards greater inflammation with HIV (1, 9). Together, these immunologic differences may increase the potential for lung damage in people with TB-HIV coinfection. If confirmed, anti-inflammatory therapies that reduce excessive lung inflammation may be particularly promising in people with TB-HIV, as found in a previous randomized trial (10).
In summary, we found that adults with pulmonary TB disease and HIV coinfection had higher levels of inflammation in both blood and lung compartments compared to those without HIV. Future studies can further explore these differences in inflammation by HIV status across the blood and lung compartments and seek to establish how these profiles may be associated with long-term outcomes and lung health after completion of TB treatment.
Data Availability
All data produced in the present study are available upon reasonable request to the authors
Funding support
This work was supported by the National Institute of Health with grants R01AI166988 [to SCA and GPB], K23AI134182 [to SCA], the Center for AIDS Research at University of Pennsylvania P30AI045008, and the Center for AIDS Research at Emory University P30AI050409.
Disclaimer
The funders had no role in the study’s design, collection, analysis, or interpretation of data, the writing of the report, or the decision to submit for publication.
Potential conflicts of interest
All authors declared no conflict of interest.
Acknowledgments
We gratefully acknowledge the staff at the Aurum Institute Tembisa Clinical Research Center, including Ms. Lindelwa Nkangane, in addition to Drs. Joshua Chandler and Robert Wallis for their scientific review and input.