The convergence of tuberculosis (TB) and type 2 diabetes mellitus (DM) threatens the life and livelihood of individuals in many resource-constrained settings. Concerns about the impact of DM on TB are particularly relevant in sub-Saharan Africa, which already bears over 80% of the world’s HIV-associated TB burden. The prevalence of DM is projected to rise steeply in the coming decades (1–3). An estimated 12.6% of the adult (age 20-79 years) sub-Sahara-African population have impaired glucose tolerance, 5.2% have type 2 diabetes, and over 50% of diabetes cases remain undiagnosed (4).

The interaction of TB and metabolic dysregulation is complex; while impaired glucose tolerance, including overt DM as well as prediabetes, increases the risk of active TB disease, the immune response to TB may also drive adverse changes in metabolic health (5–7). The increased risk of active TB in patients with DM has been attributed to multiple immunologic alterations, including monocyte and macrophage defects that contribute to delayed adaptive immune responses and altered M. tuberculosis (Mtb) specific pro- and anti-inflammatory T-helper type 1 (Th1) and Th17 responses (8, 9). Previous studies have also shown that patients with prediabetes and active TB have increased circulating levels of Th1, Th2, Th17, and regulatory cytokines (10, 11).

Although not all individuals with prediabetes go on to develop overt diabetes (12–15), the alterations in immune function in persons with prediabetes and DM that contribute to active TB disease overlap, in part, with studies linking immunologic biomarkers to the development of metabolic disease in the absence of infectious conditions. In the general population, a reduced proportion of naïve and regulatory (T_reg) CD4+ T cells (16–18), higher circulating memory CD4+ T cells (16, 19), and a shift towards pro-inflammatory Th1 and Th17 helper cells (18) was associated with prevalent diabetes. Taken together, these findings suggest a potential bidirectional relationship in which altered immunity in glucose intolerance predisposes to active TB disease, which subsequently may worsen metabolic function.

The potential for an immune response against infectious conditions to promote metabolic dysregulation has been investigated in persons with HIV (PWH) (20, 21), but there are fewer data among individuals with TB. We and others, have previously demonstrated that elevation of soluble receptor of tumor necrosis factor (TNF)-α-1 (sTNFR1) and interleukin-6 (IL-6), and increased expression of CD45RO and CD57 on CD4+ T cells, are associated with increased incidence of DM among PWH (16, 22, 23). Additionally, higher baseline frequencies of CD4+ T effector memory RA+ (T_EMRA) cells (CD45RA+ CD27-) and senescent T cells (CD4+ CD28-) have been associated with incident diabetes in PWH (21). It is unclear, however, if these factors are also associated with DM in patients with TB disease.

In this study we sought to build upon the immunometabolism model linking alterations in adaptive immunity to the development of DM to explore impaired glucose tolerance (hemoglobin A1c [HbA1c] ≥5.7%) in TB disease using longitudinal data from drug-susceptible TB patients in South Africa. We assessed a broad range of innate and adaptive immune markers among individuals undergoing treatment for active TB to identify factors associated with the persistence or worsening of glucose intolerance.

Of 62 total participants, approximately 1/3 (n = 23) had normal HbA1c levels at baseline, half (n = 33) had prediabetes, and the remainder (n = 6) were categorized as having diabetes (24). Three of the six patients with HbA1c ≥6.5% at baseline self-reported DM. The HbA1c levels of the study population are summarized in Supplementary Table 1 . The HbA1c levels for the prediabetes and diabetes groups tended to decrease following treatment, with the prediabetes group showing a significant decrease (p < 0.0001) in baseline to follow-up HbA1c values. Since the HbA1c levels varied considerably at baseline versus follow-up, we separated participants into those with any increased (n = 10) or decreased (n = 46) follow-up HbA1c relative to baseline. We did not implement a threshold to qualify an alteration as increased or decreased, rather any change was considered. Participants without change in HbA1c (n = 6) were included in the increased HbA1c group; four of the six had normal HbA1c.

Table 1 compares the population demographics/characteristics of the two groups. Neither group differed significantly by age, sex, or HIV status. All 62 participants were CMV IgG positive. Most patients (n = 51 [82%]) were cured or had treatment completion; seven had treatment failure, three did not complete treatment, and one had relapse after cure. Due to our small study sample size and since several studies have investigated the associated factors for progression from prediabetes to diabetes or reversion to normoglycemia (12–15), we focused on measuring immune-associated changes relating to HbA1c fluctuations following TB treatment and did not consider participants with overt diabetes as a separate group.

First, we assessed plasma cytokine levels of all participants to determine if changes in HbA1c following TB treatment were associated with changes in systemic markers of inflammation ( Table 2 ). Only two markers showed significant changes between study groups. CD62 P-selectin, a marker of chronic inflammation (32), was significantly higher in the patients with stable/increased HbA1c following treatment relative to those in whom HbA1c was reduced. Conversely, the anti-inflammatory cytokine IL-10 was significantly reduced in participants with stable/increased HbA1c but remained unchanged in patients with decreased HbA1c. Although no significant differences were observed for the other soluble factors measured, including proinflammatory markers such as IL-6 and IL-1b, these data suggested an association between increasing HbA1c and a more proinflammatory state following TB treatment.

To determine if stable/increased HbA1c levels were associated with differences in T cell immunity, PBMC samples were stimulated as described and an unbiased clustering approach used to identify CD4+ or CD8+ T cell phenotypes of interest ( Figure 1A ). The main cytokine producing CD4+ populations, clusters 2-7 ( Figure 1B ), all grouped together and displayed either an effector memory (CD45RO+ CCR7-) or a TEMRA (CD45RO- CCR7-) phenotype. This approach identified a PMA-stimulated IL-17+ TNF-α+ CD4+ cluster (cluster 6) which was positively associated with HbA1c levels ( Figures 1C, D ). At follow-up after TB treatment, the capacity of CD4+ T cells to produce IL-17 was increased in patients with stable/increased HbA1c. This population has an effector memory phenotype (CD45RO+ CCR7-) and comprises about 0.6% of total CD4+ T cells. In addition, Mtb-specific T cells, which produced IL-17 but not TNF-α+ following stimulation with MTB300, were also significantly elevated in participants who had stable/increased HbA1c. These cells had a TEMRA (CD45RO- CCR7-) phenotype. To confirm these differences identified using an unbiased approach, we gated on total cytokine production for CD4+ T cells using conventional Boolean gating ( Table 3 ; Supplementary Figure 1 ). Despite seeing no difference in plasma TNF-α, this revealed a significant increase in TNF-α production from CD4+ T cells stimulated with MTB300 antigen in patients with stable/increased HbA1c. Together these data suggest that stable/increased HbA1c following TB treatment was associated with elevated Th1 and Th17 CD4+ responses. In contrast, the capacity for Th2 responses (based on the production of IL-4/-13) was reduced in participants with stable/increased HbA1c, reaching significance at follow-up with PMA stimulation ( Supplementary Table 4 ).

An additional phenotyping flow panel was used, and unbiased clustering revealed distinct clusters suggestive of activated and exhausted T cells (based on the expression of HLA-DR, PD-1, CD57 and CX3CR1) ( Figures 2A, B ). Of these, cluster 6, distinguished by expression of CX3CR1 in the absence of HLA-DR or CD57, showed a positive association with stable/increased HbA1c ( Figure 2C ). Furthermore, looking at total CX3CR1+ CD4+ T cells, their frequency was higher at baseline and follow-up in the patients with stable/increased HbA1c (~1.4 fold higher). In addition, cluster 8 (CD57+) almost doubled in frequency from baseline to follow-up for patients with stable/increased HbA1c, but this did not reach statistical significance.

Examining CD8+ T cell responses in the same way we found that a PMA stimulated IFNγ+ TNF-α+ effector memory CD8+ T cell population (cluster 9) was upregulated in patients with stable/increased HbA1c at follow-up ( Figure 3 ). As expected, the IL-17, IL-4 and IL-13 expression were confined to CD4+ T cells, with little or no detection in CD8+ populations. From the phenotyping panel ( Figure 4 ), cluster 2, consistent with CD8+ TEMRA CD57+ population, was lower in patients with stable/increased HbA1c at both baseline and follow-up but reached significance at follow-up ( Figure 4C ). A similar trend, although not significant, was observed for cluster 6, which expressed markers of activation and proliferation (HLA-DR+ Ki67+).

In our drug-susceptible TB study cohort with high proportion of PWH, we investigated the immune factors associated with persistent glucose intolerance following treatment of active disease. Longitudinal sampling allowed us to group patients based on stable/increased HbA1c or decreased HbA1c. Stable/increased HbA1c one year after TB treatment initiation was associated with a pro-inflammatory state reflected by elevated plasma cytokines, Th1 (CX3CR1+) and Th17 CD4+ T cell responses.

Six patients had an HbA1c ≥6.5% at baseline, three of whom received anti-diabetic medications, though they did not start these medications during follow up. Despite a relatively low proportion of patients with DM in our cohort, over half (56%) had HbA1c levels consistent with prediabetes at baseline. We found that most patients had resolution of hyperglycemia after TB treatment, consistent with well-documented transient hyperglycemia (33). It is unclear whether the hyperglycemia observed in our study and others increases long-term risks for development of DM (34).

Although most participants did not have baseline HbA1c levels indicative of overt DM, and most had resolution of hyperglycemia, those with increased HbA1c levels tended to display a more pro-inflammatory state characterized by elevated CD62 P-selectin and depressed IL-10 plasma levels. Although less pronounced, likely due to the small sample size, this is in keeping with similar trends observed by Randeria et al. who compared type 2 diabetics (HIV and TB negative) to healthy controls (35). HIV and TB co-infection as well as the effect of ART may have influenced our findings. Nonetheless we observed elevated soluble CD62 P-selectin, which serves as an indicator of vascular inflammation and is characteristic of type 2 diabetes (36–38). Several studies have described either an increased expression of IL-10 in plasma of persons with type 2 diabetes (35) or unchanged/reduced levels (39, 40) relative to healthy controls. Interestingly, glucose intolerance reduces IL-10 signaling efficiency, resulting in increased TNF-α production (39).

Although we did not observe a difference in TNF-α, we did detect reduced IL-10 plasma levels. Reduced IL-10 signaling has been associated with type 2 diabetes (40). Together the lower IL-10 levels and reduced signaling tie in well with a pro-inflammatory T cell response. In this regard, participants with HbA1c levels that increased or remained stable over time had an increased capacity of CD4+ T cells to produce IL-17 and TNFα (following PMA stimulation), and for Mtb-specific CD4+ T cells to produce IL-17 alone following stimulation with the TB-antigen pool MTB300 (41). Recent data from a human TST challenge model suggest that TB-specific Th17 T cell responses are associated with TB pathogenesis (42). Although most participants had successful treatment and were culture negative at the end of TB therapy, it is possible that incomplete Mtb killing may occur within the lungs of some TB patients, as demonstrated in human studies using direct sampling of the lung environment (43). If so, then residual infection might be expected to prevent the resolution of HbA1c and be associated with persistent inflammation and pathogenic IL-17 T cell responses. Regardless, dysglycemia is associated with decreased Th1 and Th17 responses to Mtb antigens in individuals with latent TB infection (44, 45), suggesting a different relationship following TB treatment. Interestingly, the depression of these responses was linked to elevated IL-10, which was significantly lower in individuals following treatment who had stable or increased HbA1c levels, supporting this hypothesis.

In addition, we observed an association between CD4+ T cell expression of CX3CR1, a marker associated with Th1 effector populations (46, 47), and stable/increased HbA1c levels. Lau et al. also found DM was associated with an increase in CX3CR1 expressing T cells and a reduced Th2 response (48). The CX3CR1-fractalkine axis mediates the migration of CX3CR1 expressing lymphocytes to inflamed tissue endothelium (46, 49). These T cells often have anti-viral and cytotoxic properties and are associated with CMV and HIV infection (31, 50). All participants in this study had CMV reactive plasma and 69% of participants in the stable/increased HbA1c group were HIV positive, therefore this increased CX3CR1+ CD4+ T cell frequency is also consistent with their anti-viral properties. Nonetheless, CX3CR1 expression on CD4+ T cells associated positively with stable/increased HbA1c. This, in turn, is consistent with this phenotype’s association with the progression of metabolic disease (20).

Our study had limitations. First, our sample size was small. Despite the small number of patients with DM at baseline, we were able to demonstrate the association of stable/increased HbA1c with elevated Th1 and Th17 CD4+ responses, providing direction for further study. Second, we had a single follow-up measurement of HbA1c and CD4+ cells 12-15 months after TB treatment initiation. Although this timeframe allowed time for a complete course of treatment, it remains unclear what risk, if any, is conferred by the dysregulation we observed in those whose HbA1c remained stable or increased. Finally, our investigations were exploratory and need to be confirmed in additional contexts and populations.

Identifying markers that associate with glucose intolerance could help identify metabolically fragile individuals who need closer monitoring/treatment. Monitoring the pro-inflammatory T cell response in hyperglycemic patients after TB treatment could prove useful in this regard. Further work investigating the mechanisms linking these immune pathways to metabolic dysregulation and poor clinical outcomes are needed.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

This study was approved by the Vanderbilt Institutional Review Board (IRB#191176). Informed consent was obtained under the Biomedical Research Ethics Committee (BREC) at the University of Kwazulu-Natal (UKZN) Ref No: BE423/19, and the RePORT South Africa study (IRB#160918) at Vanderbilt. The patients/participants provided their written informed consent to participate in this study.

RK, CW and JS share co-first authorship. RK conducted experimental work, analysis and wrote the manuscript. CW contributed to analysis and writing. JS contributed to analysis and writing. PR contributed to statistical analysis and writing. FM, FK and TS contributed to study design, administration and writing. JK contributed to study design, analysis and writing. AL contributed to study design and writing. YV contributed to study design, funding, analysis and writing. All authors contributed to the article and approved the submitted version.