Gene expression in TB disease measured from the periphery is different from the site of infection

Although the incidence and mortality of tuberculosis (TB) is decreasing, the rate of decrease is not enough to attain the WHO End TB strategy to reduce the global burden of TB by 2035. Furthermore, the impact of the COVID-19 pandemic in TB disease may result in an additional half a million deaths potentially increasing mortality to the 2010 levels [1]. The diagnosis of TB relies on detection of Mycobacterium tuberculosis (Mtb) in sputum either by smear microscopy or culture. Smear has low sensitivity whilst culture, the gold standard, has a sensitivity ranging from 80% to 93% and a specificity of 98% [2]. However, due to the slow growth of the bacteria, it may take up to six weeks for a negative result and requires extensive infrastructure. GeneXpert MTB/RIF (Cepheid) is a PCR based test that allows rapid diagnosis in tertiary level clinics but also requires extensive infrastructure.

The development of tests based on host-markers present in the blood has been of interest recently. Transcriptomic gene signatures can differentiate TB disease from Mtb infection and other respiratory diseases suggesting their potential use as diagnostic and prognostic markers. While large numbers of genes in these signatures have hampered their translation to clinical tests; there is an overlap of certain differentially expressed genes in different signatures suggesting they may be predominant.

Some of these genes, including Type I IFN stimulated genes such as BATF2, GBP5, ANKRD22 and CDC42 are conserved in gene expression signatures in humans, non-human primates and mice, expressed much higher in TB disease relative to healthy controls [3,4]. However, genes were evaluated in whole blood based on empiric evidence that gene expression in the periphery is indistinguishable from gene expression observed at the site of infection. Thus, to improve blood signatures and to understand mechanism of gene expression, it's important to study the direct relationship between the lung and whole blood. Acquisition of lung tissue and granuloma from humans is difficult due to its invasive nature, however an alternative to lung tissue is cells collected from the bronchoalveolar lavage fluid (BALF). A recent study observed different gene expression patterns between blood and cells collected from the bronchoalveolar fluid in both HIV-infected and uninfected persons with or without TB [5]. However, it is unknown if this holds true in patients from West Africa. Thus, we measured gene expression in whole blood and BALF from presumptive TB patients based on clinical triage with at least one TB symptom, and/or with a history of prior TB to determine if blood can act as a surrogate for BAL in Gambians. Differential gene expression of BATF2 (a regulator of inflammation), GBP5 (an activator of NLRP3 inflammasome which plays a role in innate immunity), ANKRD22 (expressed close to TB disease) and CDC42 (encoded by GTPases, which controls diverse cellular function and cell division) (Table 1 ), were measured in 11 patients with presumptive TB by real time (quantitative) polymerase chain reaction (qPCR). Genes were normalised against a housekeeping gene (B2M) generating fold chain expression values. Wilcoxon sign rank test (paired test) and Kruskal-Wallis test with Dunn's post-test comparison were used to determine differences between two and three groups, respectively. Non-parametric Spearman correlation was used to determine an association between whole blood and BAL. P-values less than 0.05 were considered statistically significant.

Table 1.

TaqMan primer probes.

TaqMan Primer Probe	Assay ID	Role
Guanylate Binding Protein 5 (GBP5)	Hs00369472_ml	Increased in progressors and during TB disease
Cell division control 42 (CDC42)	Hs00970458_ml	Increased in TB disease
Basic leucine zipper transcription factor 2 (BATF2)	Hs00912736_ml	Increased in TB disease and during progression to TB
Ankyrin repeat domain 22 (ANKRD22)	Hs00944018_ml	Higher during TB disease and in progression to TB
Beta-2 Microglobulin (B2M)	Hs99999907_ml	Housekeeping gene

Gene expression was higher in BAL than whole blood for the housekeeping gene and genes of interest (data not shown). Following normalisation, no difference in gene expression was observed between whole blood and BALF (Figure, A – D), however, we did not observe any correlation in gene expression between whole blood and BAL samples (E − H) as previously shown [5].

We next analysed the same genes (in the blood only), to compare their expression as paired samples in TB patients pre- and post-treatment (n = 10) compared to latently TB infected household contacts (n = 10). Of the four genes, only GBP5, wasn't differentially expressed between ATB and LTBI (Figure, L). CDC42, was significantly higher in ATB compared to LTBI (p = 0.04) but no difference was seen at diagnosis and end of treatment (Figure, K). BATF2, previously shown to differentiate ATB from healthy controls [4], was significantly higher in ATB compared to both post-treatment (p = 0.0082) and LTBI (p = 0.003) (Figure, J). No difference was seen between post-treatment and LTBI levels. ANKRD22, was expressed at low levels in our cohort as observed in the BAL and whole blood cohort but was significantly higher in ATB compared to LTBI only (p = 0.01) (Fig. 1 ).

Fig. 1.

Differential gene expression in whole blood and BALF and at the different stages of TB disease. A-D): Differences in fold change expression for genes of interest relative to housekeeping gene between whole blood and BAL, shown in Log10. Lines represent medians, dots represent individual responses. Only significant p-values (p < 0.05) are shown, p-values calculated using Kruskal-Wallis and Dunn's multiple comparison test. E-H): Correlation of fold change expression for genes of interest relative to housekeeping gene between whole blood and BAL, shown in Log10. I-L): Differences in fold change expression for genes of interest relative to housekeeping gene between pre and post treatment samples from cases and healthy controls shown in Log10. Lines represent medians, dots represent individual responses. Only significant p-values (p < 0.05) are shown, p-values calculated using Kruskal-Wallis and Dunn's multiple comparison test.

To conclude, we have shown that Type I IFN genes measured by qPCR are highly expressed in BAL cells of persons with presumptive TB after bronchoscopy. However, we observed that gene expression measured in whole blood is not indicative of expression in the lung as measured from BAL cells. These genes can, however, differentiate between active TB disease and latent infection in whole blood collected from Gambians.

Author contributions

FD and JS designed the study. OO, ST, AJ, MD, BF, and RJ are in the clinical team, enrolled participants, performed clinical assessments, phlebotomy and bronchoscopy. CNM, AB, EGC and DJ processed samples and stored them. SD was the data manager for the study. DJ and FD designed and performed the experiments. DJ, FD and JS wrote the manuscript. All authors reviewed the article.

Appendix A. Supplementary data

The following is the Supplementary data to this article: