Comorbidity between counnunicable and non-communicable diseases : the example of the dual burden of tuberculosis and diabetes in Dar es Salaam, Tanzania

Background
Although recognized for centuries, the association between tuberculosis (TB) and diabetes mellitus (DM) was forgotten with the discovery of efficient treatments. In the last decade, the prevalence of DM has dramatically increased, particularly in low- and middle-income countries experiencing a high burden of TB, leading to a new interest in this association. DM increases TB risk while TB, as an infectious disease, leads to hyperglycemia. The relationship between TB and DM has been poorly studied in Sub-Saharan Africa, where the high incidence of TB is associated with HIV infection. Concentration of vitamin D is inversely associated with TB and DM, and it has been suggested that low vitamin D could mediate some of the association between TB and DM. DM affects the immune response to TB, but the precise mechanisms underlying this association are not clear.
To address this issue of high public health relevance, we undertook a project on the association between TB, DM and HIV in Tanzania. The project had three major components:
(1) Assessing the association of TB and its outcome with the presence and persistence of hyperglycemia in Tanzania, using three different DM screening tests.
(2) Describing the association between vitamin D, TB and DM.
(3) Studying the immunological features underlying TB and DM comorbidity in sub-Saharan Africa and testing the hypothesis of delayed adaptive immune response with increasing glycemia.
The overall aim of the project was to improve knowledge on the dynamic interaction between TB and DM in an African setting with high HIV prevalence by integrating a longitudinal component into the case-control study.
Methods
A case-control study with longitudinal follow-up of cases was conducted in Dar es Salaam. Consecutive adults with new active TB were included and followed up for five months after the start of anti-TB treatment. Healthy controls, matched by age and sex to TB cases, were recruited among volunteering adults accompanying patients to the outpatient departments of the same hospitals. Exclusion criteria were a biological relationship to TB case, TB history, symptoms or signs of TB, other acute infection or major trauma within the last three months. All underwent 25-hydroxyvitamin D (25(OH)D) measurement and DM screening tests (fasting glucose (FCG), 2-hour capillary glucose after standard oral glucose tolerance test (2h-CG) and glycated hemoglobin (HbA1c)) at enrolment and TB patients were again tested after five months of TB treatment. Data on the outcome of TB (treatment failure, death, lost to follow-up) were collected.
For the nested immunological study, four groups of HIV negative patients were included: i) active TB without DM, ii) active TB with DM, iii) latent TB patients without DM and iv) latent TB patients with DM. Latent TB patients were selected among the healthy volunteering adults, as well as among diabetic patients attending the DM clinic in the participating hospitals. Exclusion criteria for groups iii and iv were past TB history and symptoms or signs of active TB. Peripheral blood mononuclear cells were stimulated with Mycobacterium tuberculosis (Mtb)-specific peptide pools and live Mycobacterium bovis BCG and then analysed by polychromatic flow cytometry for Th1, Th2, Th9 and Th17 cytokine production. Cell culture supernatants were analysed by Luminex® for 34 cytokines and chemokines.
Findings
At enrolement, DM prevalence was significantly higher among TB patients (N=539; FCG>7mmol/L: 4.5%, 2-hCG>11mmol/L: 6.8% and HbA1c>6.5%: 9.3%) compared to controls (N=496; 1.2%, 3.1% and 2.2%). However, the association between hyperglycemia and TB disappeared after TB treatment (aOR(95% CI) at enrolment vs follow-up: FCG 9.6(3.7-24.7) vs 2.4(0.7-8.7); 2-hCG 6.6(4.0-11.1) vs 1.6(0.8-2.9); HbA1c 4.2(2.9-6.0) vs 1.4(0.9-2.0)). FCG hyperglycemia at enrolment was associated with TB treatment failure or death (aOR(95%CI) 3.3(1.2-9.3).
The prevalence of 25(OH)D insufficiency (25(OH)D<75nmol/l) was not statistically different between TB patients and controls (25.8% versus 31.0%; p=0.22). But the association between 25(OH)D insufficiency and TB was modified by hyperglycemia (pinteraction=0.01). Patient with vitamin D insufficiency were only at higher risk for TB in the presence of underlying hyperglycemia. The OR (95%CI) for TB risk in patients with vitamin D insufficiency and hyperglycemia was 4.94(1.16-21.0) versus 0.68(0.39-1.17) for patients with vitamin D insufficiency and normoglycemia where normoglycemia and normal vitamin D were the reference category.
Patients with active TB and DM had a lower frequency of INF-γ CD4+ T cells and a lower proportion of CD4+ T cells producing both TNF-α and IFN-γ after live M. bovis BCG but not after Mtb-specific peptide pool stimulation, compared to normoglycemic TB patients. A negative correlation between INF-γ or TNF-α CD4+ T cell frequency and increasing glycemia was observed in the context of live M. bovis BCG stimulation only.
Conclusions
Transient hyperglycemia is frequent during TB, and DM needs confirmation after TB treatment. However, DM screening at TB diagnosis gives the opportunity to detect patients at risk of adverse outcome.
25(OH)D insufficiency seams to increase the risk of TB only if associated with hyperglycemia. DM patients living in high TB burden settings might benefit from preventive vitamin D supplementation.
The immunological findings suggest that DM might affect Mtb-specific CD4+ T cell immune responses at the level of reduced antigen processing and presentation, a defect that could be compensated by metformin.
The results of the study are of public health and clinical utility. First, they lend support to the integration of care between TB and DM programs. Second, they imply that, at the time of TB diagnosis, patients should be screened for hyperglycemia using cost-effective fasting glucose tests. Treatment of hyperglycemia should be initiated to improve TB outcome. Third, before initiation of long-term DM treatment, DM diagnosis must be confirmed after the resolution of TB. Finally, in the absence of evidence for a strong contribution of DM to TB risk in this African setting with high HIV prevalence, DM patients should not be screened for TB with expensive test. DM physicians and patients should rather be trained for recognizing TB symptoms and signs as a cost-effective way to recognize TB early.