Machine learning reveals Mycobacterium tuberculosis genotypes and anatomic disease site impacts drug resistance and disease transmission among patients with proven extra-pulmonary tuberculosis

Abstract

Background Genotypic characterization of Mycobacteria tuberculosis ( Mtb ) isolates in pulmonary tuberculosis is well documented; however, there is dearth of information on extrapulmonary tuberculosis (EPTB). Here, we investigate Mtb drug resistance and transmission patterns in EPTB patients diagnosed and treated in the Tshwane metropolitan, South Africa. Methods Consecutive Mtb culture-positive non-pulmonary isolates underwent mycobacterial genotyping and were assigned to phylogenetic lineages and transmission clusters based on spoligotypes. The Genotype MTBDR plus assay was used for isoniazid and rifampin susceptibility testing. Ensembles of machine learning (ML) algorithms were used to identify clinically meaningful patterns in data. We computed odds ratio (OR), attributable risk (AR) and their corresponding 95% confidence intervals (CI). Based on ML outputs, we examined factors associated with EPTB sites, drug resistance and clustering for the period between July 1 st , 2014 and January 31 st , 2015. Results A total of 75 isolates were identified during study period: we excluded 5 (7%) Mycobacterium bovis isolates. The overall proven EPTB incidence of 4.43 (95% CI: 3.72-5.23) per 100,000 population per year is lower than prior reports. However, the frequency distribution across most frequently encountered EPTB disease sites, including pleura, peritoneum, meninges and skin, were consistent with those reports. The largest cluster comprised of 25 (36%) Mtb strains belonged to East Asian lineage. East Asian lineage was significantly more likely to occur within a chain of EPTB transmission when compared to Euro-American and East-African Indian lineages: OR= 10.11 (95% CI: 1.56-116). Lymphadenitis, meningitis and skin TB, were significantly more likely to be associated with drug resistance: OR=12.69 (95% CI: 1.82-141.60) and AR = 0.25 (95% CI: 0.06-0.43) when compared with other EPTB sites, which suggests that poor rifampin penetration might be a contributing factor. Conclusions Majority of Mtb strains circulating in the city of Tshwane metropolis belongs to East Asian, Euro-American and East-African Indian lineages, and each of these are likely to be clustered, suggesting on-going EPTB disease transmission. On the other hand, 25% of the drug resistance is attributable to sanctuary EPTB sites notorious for poor rifampin penetration, suggesting role for poor drug dosing.