Inflammation and dyslipidaemia in combined diabetes and tuberculosis; a cohort study

Abstract

Diabetes mellitus (DM) increases tuberculosis (TB) susceptibility and worsens outcomes. Since inflammation and lipid metabolism are implicated in both diseases, we examined if combined TB and DM (TB-DM) increases inflammation or dyslipidaemia. In plasma from individuals with DM (n = 96), TB (n = 93), and TB-DM (n = 91), we measured 92 inflammatory proteins and 250 primarily lipid-related metabolites, repeating measurements after two months of TB treatment. Inflammation was primarily driven by TB, but higher in TB-DM. In TB-DM, the proteins osteoprotegerin (OPG), signaling lymphocytic activation molecule (SLAMF1), adenosine deaminase (ADA), interleukin-10 receptor subunit beta (IL-10RB), and tumor necrosis factor receptor superfamily member 9 (TNFSR9) were differentially abundant, and IL-17A/C predicted treatment failure. Disease severity correlated with inflammation and dyslipidaemia. Inflammation decreased with TB treatment, both in TB and TB-DM. Dyslipidaemia was primarily driven by DM, but more pro-atherogenic in TB-DM, with elevated VLDL and apolipoprotein B (ApoB). Despite TB treatment, pro-atherogenicity persisted. Stronger inflammation and dyslipidaemia may account for worse disease outcomes in TB-DM and warrant further action to prevent cardiovascular events.

Keywords: Endocrinology; Health sciences; Immunology; Medical microbiology; Medical specialty; Medicine.

Graphical abstract

Figure 1

Inflammation at baseline Comparison of circulating inflammatory markers of individuals with DM (n = 96), TB (n = 93), and TB-DM (n = 91) at baseline. (A) Principal-component analysis (PCA) plot shows the first two principal components derived from individuals with DM (yellow), TB (blue), and TB-DM (pink). (B) Heatmap representing the z-scores of the median of the three groups for each measured circulating inflammatory marker. Higher and lower Z scores are depicted in red and blue, respectively. (C) Radial Volcano plot showing a cross-section of a three-dimensional Volcano plot comparing circulating inflammatory markers between the three groups. Circulating inflammatory markers, which are nonsignificant (gray), only significantly higher in DM (red), TB (not applicable), and TB-DM (yellow) and overlapping significance between DM and TB-DM (green) as well as TB and TB-DM (turquoise) are shown. The z axis is plotted as -log₁₀(adjusted p value) and the polar coordinates display the relative mean level of the measured marker. Statistical testing was performed by Wilcoxon rank-sum test (two-sided) between two of the three groups and Kruskal-Wallis test to compare all three groups.

Figure 2

Inflammation after two months of TB treatment Paired changes of circulating inflammatory markers after two months of TB treatment in individuals with TB (n = 31) and TB-DM (n = 33) are displayed. (A) Principal-component analysis plot showing principal component one and two for individuals with DM (yellow), TB (blue), and TB-DM (pink) at baseline and for TB (blue, triangles) and TB-DM (pink, triangles) after two months of TB treatment. (B) Combined Volcano plot displaying the Log₂ fold change between baseline and after two months of TB treatment for TB and TB-DM. Significant changes between two months TB treatment and baseline in both groups (green), in TB alone (blue), in TB-DM alone (pink), and nonsignificant changes (gray) are displayed. Statistical testing was done by paired Wilcoxon ranked-sum test (adjusted p value <0.05). (C) Boxplots showing the median plus interquartile ranges of normalized protein expression (NPX) for 4E-BP1, IFNy, IL-17C, IL-18R1, and SCF as examples for changing marker levels after two months of TB treatment. Statistical testing was done by paired Wilcoxon ranked-sum test (adjusted p value <0.05). ∗ adj. p < 0.05, ∗∗ adj. p < 0.01, ∗∗∗ adj. p < 0.001.

Figure 3

Associations of inflammatory proteins with clinical characteristics and treatment outcome (A) Linear regressions and absolute effect sizes representing the association between baseline circulating inflammatory markers and age, sex, BMI, HbA1c, hemoglobin, and Timika score for individuals with DM (n = 96), TB (n = 93), and TB-DM (n = 91). (B) Linear regressions and absolute effect sizes representing the association between circulating inflammatory markers after two months of TB treatment and a positive sputum culture result after six months of TB treatment representing treatment failure in individuals with TB-DM (negative sputum culture at month two [n = 28], positive sputum culture at month two [n = 3]). (A and B) Negative and positive associations are depicted in blue and red, respectively (FDR <0.05). The absolute effect size is depicted by the size of the square. All variables were tested independently by linear regression.

Figure 4

Lipids at baseline Comparison of lipid intermediates and metabolic markers of individuals with DM (n = 93), TB (n = 91), and TB-DM (n = 83) at baseline. (A) Principal Component Analysis (PCA) plot shows the first two principal components derived from individuals with DM (yellow), TB (blue), and TB-DM (pink). (B) Heatmap representing the Z scores of the median of the three groups for each measured marker. Higher and lower Z scores are depicted in red and blue, respectively. (C) Radial Volcano plot showing a cross-section of a three-dimensional Volcano plot comparing Lipid intermediates and metabolic markers between the three groups. Markers, which are nonsignificant (gray), only significantly higher in DM (red) or TB-DM (yellow) and overlapping significance between DM and TB-DM (green) as well as TB and TB-DM (turquoise) are shown. The z axis is plotted as -log₁₀(adjusted p value) and the polar coordinates display the relative mean level of the measured marker. Statistical testing was performed by Wilcoxon rank-sum test (two-sided) between two of the three groups and Kruskal-Wallis test to compare all three groups.

Figure 5

Lipids after two months of TB treatment Paired changes of lipid intermediates and metabolic markers after two months of TB treatment in individuals with TB (n = 32) and TB-DM (n = 31) are displayed. (A) Principal-component analysis plot showing principal component one and two for individuals with DM (yellow), TB (blue), and TB-DM (pink) at baseline and for TB (blue, triangles) and TB-DM (pink, triangles) after two months of TB treatment. (B) Combined Volcano plot displaying the Log₂ fold change between baseline and after two months of TB treatment for TB and TB-DM. Significant changes between two months TB treatment and baseline in both groups (green), in TB alone (blue), in TB-DM alone (pink), and nonsignificant changes (gray) are displayed. Statistical testing was done by paired Wilcoxon ranked-sum test (adjusted p value <0.05). (C) Boxplots showing the median plus interquartile ranges of the plasma concentration for ApoB (g/L), GlycA (mmol/L), HDL-C (mmol/L), HDL-TG (mmol/L), and LDL-C (mmol/L) as examples for changing marker levels after two months of TB treatment. Statistical testing was done by paired Wilcoxon ranked-sum test (adjusted p value <0.05). ∗ adj. p < 0.05, ∗∗ adj. p < 0.01, ∗∗∗ adj. p < 0.001.

Figure 6

Associations of lipid markers with patient characteristics Linear regressions and absolute effect sizes representing the association between baseline circulating lipid intermediates and metabolic markers and age, sex, BMI, HbA1c, hemoglobin, and Timika score for individuals with DM (n = 93), TB (n = 91), and TB-DM (n = 83). Negative and positive associations are depicted in blue and red, respectively (FDR <0.05). The absolute effect size is depicted by the size of the square. All variables were tested independently by linear regression.

Figure 7

Correlation between inflammatory and lipid markers Heatmap displaying the spearman’s rank correlation coefficient between absolute lipid intermediate concentrations and normalized protein expression (NPX) values of circulating inflammatory markers for individuals with (A) DM (n = 92), (B) TB (n = 90), and (C) TB-DM (n = 82). Positive and negative correlation coefficient are depicted by red and blue, respectively (FDR <0.05).