Background Diabetes mellitus (DM) maybe a risk factor for tuberculosis (TB) and negatively affect outcome of treatment. We investigated the impact of DM on TB treatment outcome in a longitudinal study.

Methods The diabetic status of all microbiologically confirmed TB patients was determined at baseline (t0). The DM group consisted of known DM and on anti-diabetics (TBDMt) and newly diagnosed DM patients who were monitored by clinicians for three months without DM treatment (TBDMnt). Bacterial clearance at days 0, 7, 14, 28 and 56 were analysed by molecular bacterial load assay (MBLA) and HbA1c levels at three (t1), and six (t2) months during TB treatment, and 3 months post-treatment (t3). Clinical examination including X-ray was done.

Results A total of 559 TB patients were followed, 49 (8.8%) were diabetic with 36 (6.4%) TBDMt and 13 (2.4%) TBDMnt. The HbA1c of the TBDMt showed significant decrease in median-HbA1c from t0 to t3 (p=0.029) but still hyperglycemic. The TBDMnt cohort showed a significant decline in median-HbA1c from t0 to t1 (p=0.006), and remained normoglycemic. The TBDMt cohort had more infiltrations in the lower lung fields than the TB-only cohort (p=0.02). Analysis of 354 serial sputa collected from 59 cases (42 TB-only and 17 TBDMt cases) showed that the average bacterial load of the TB-only cohort at diagnosis was significantly higher than TBDMt cohort (p=0.03). However, after 56 days of TB treatment, the bacillary load of TBDMt cohort was significantly higher (p=0.04) using TB-MBLA. Time of sputum conversion from positive to negative as measured by TB-MBLA was shorter in TB-only participants (66 days) than TBDMt participants (88 days). Genotyping of mycobacterial isolates showed that Mycobacterium africanum Lineage 6 was associated with TBDMt cohort (p=0.023).

Conclusion Our findings suggest delayed mycobacterial clearance among DM cohorts and DM may predispose individuals to TB caused by distinct lineages.