Metabolomics Strategy Assisted by Transcriptomics Analysis to Identify Potential Biomarkers Associated with Tuberculosis

Figures & data

Figure 1 An overview workflow of the comprehensive analysis of metabolomics and transcriptomics in tuberculosis.

Table 1 Characteristics (n = 90) for TB Cases, LTBI and HC Subjects*

	HC (n=30)	LTBI (n=30)	TB (n=30)	P
Age	47.5 (27–61)	44.5 (27–60)	54.5 (28–65)	0.721
Sex, male	25 (83.3)	17(56.7)	25 (83.3)	0.027
BMI	22.7 (21.3–25.3)	22.5 (20.8–23.9)	22.2 (18.0–23.3)	0.095
Co-morbidities
Hypertension (Yes)	5 (16.7)	3 (10.0)	3(10.0)	0.78
Diabetes mellitus (Yes)	1 (3.3)	1 (3.3)	6 (20.0)	0.119
Other (Yes)	6 (20.0)	1 (3.3)	7 (23.3)	0.076
Pulmonary TB diagnosis
Sputum AFB smear (positive)	–	–	20 (66.7)
Sputum AFB culture (positive)	–	–	1 (3.3)
Smoking last 6 months (yes)	12 (40.0)	6 (20.0)	7 (23.3)	0.202
Alcohol past 6 months (yes)	2 (6.67)	0 (0)	4 (13.3)	0.159

Note: *Data are shown as counts and percentages except for age and BMI (summarized using mean and range).

Abbreviations: TB, tuberculosis; LTBI, latent tuberculosis infection; HC, healthy control; BMI, body mass index; AFB, acid-fast bacillus.

Figure 2 Multivariate statistical analyses of the different clinical groups in the study. (A) Principal component analysis (PCA) score plots of the TB patients (TB, red), latently infected tuberculosis (LTBI, green) and healthy controls (HC, blue) serum samples. Orthogonal partial least squares-discriminate analysis (OPLS-DA) score plots for the comparison between (B) HC (blue) and LTBI (green) (R²Y= 0.937, Q²= 0.135), (C) TB (red) and HC (blue) (R²Y= 0.895, Q²= 0.709) and (D) TB (red) and LTBI (green) (R²Y= 0.952, Q²= 0.776).

Figure 3 Volcanic map of differential metabolites. For the LTBI/HC group (A) and LTBI/TB (B) and TB/HC group (C). The abscissa represented the fold change of the group compared to each substance (take the base 2 logarithm), the ordinate represented the P-value (take the base 10 logarithm), and the scatter size represented the VIP value of the OPLS-DA model. The larger the scattering, the larger the VIP value. The scatter color represented the final screening result. Significantly up-regulated metabolites are shown in red (FC ≥1, P < 0.05), significantly down-regulated metabolites are shown in blue (FC ≤-1, P < 0.05), and non-significant differential metabolites are shown in gray. (D) The Venn diagrams show the overlap between different health statuses for metabolites with significant abundance differences (VIP ≥1 and P < 0.05) between LTBI/HC group, LTBI/TB, and TB/HC group.

Table 2 Coefficient and Cut-off Using Metabolites Biomarkers Distinguishing Between TB and HC Subjects

	TB vs HC
	Coefficient	AUC of a Single Metabolite
(Intercept)	3.10
Heptadecanoic acid	−450.99	0.73
Unknown … 71	−1239.80	0.81
Indole acetaldehyde	−91.87	0.69
5-hydroxyindoleacetic acid	7727.07	0.69
Iso leucyl-isoleucine	7809.13	0.71
5-ethyl-2,4-dimethyloxazole	−3007.57	0.66
2-hydroxycaproic acid	−47,620.14	0.76

Figure 4 Receiver operating characteristic model of the single metabolites included in the LASSO regression equation. The ROC curves of each metabolite showed a moderate distinguishing efficiency in TB vs HC group.

Figure 5 Metabolite set enrichment analysis (MSEA) of differentiating metabolites from TB patients and HC subjects. The scatter shows the most altered metabolic pathways with fold enrichment higher than 1 (increasing P-value coloured from red to white).

Table 3 Nine Genes Enriched in 3 Pathways

Gene	Function	Pathway	Trend (TB/HC)*	AUC
SDS	Serine dehydratase	Valine, leucine, and isoleucine biosynthesis	↓	0.981 (0.930–1.000)
SDSL	Serine dehydratase like	Valine, leucine, and isoleucine biosynthesis	↓	1.000 (1.000–1.000)
BCAT1	Branched chain amino acid transaminase 1	Valine, leucine, and isoleucine biosynthesis	↓	1.000 (1.000–1.000)
BCAT2	Branched chain amino acid transaminase 2	Valine, leucine, and isoleucine biosynthesis	↑	0.852 (0.635–1.000)
GLYAT	Glycine-N-acyltransferase	Phenylalanine metabolism	↑	1.000 (1.000–1.000)
ALDH3B1	Aldehyde dehydrogenase 3 family member B1	Phenylalanine metabolism	↑	0.981 (0.930–1.000)
IL4I1	Interleukin 4 induced 1	Phenylalanine metabolism; Phenylalanine, tyrosine, and tryptophan biosynthesis	↓	1.000 (1.000–1.000)
GOT1	Glutamic-oxaloacetic transaminase 1	Phenylalanine metabolism	↓	1.000 (1.000–1.000)
MIF	Macrophage migration inhibitory factor	Phenylalanine metabolism	↑	0.963(0.877–1.000)

Notes: *The trend of marker levels in the active TB group. ↑ and ↓ indicate increased and decreased levels, respectively, compared with the healthy group.

Abbreviation: TB, tuberculosis.