In South Africa, multi- and extensively drug-resistant tuberculosis (MDR-TB and XDR-TB) have unfavourable outcomes in ~50-75% of cases, and despite comprising less than 2% of the total TB burden, consumes ~60% of the annual TB drug budget. Epidemiological modeling studies (1, 2) have shown that the disruption of person-to-person transmission would drastically limit the DR-TB epidemic. However, the provision of universal infection control facilities is unrealistic given the sheer burden of disease. It is well established that TB patients who have microscopically-detectable acid-fast bacilli in their sputum are more infectious that those who do not (3), however, epidemiological and experimental studies have shown that there is a spectrum of infectiousness, and that only a minority of such patients (10-20%) are responsible for the majority of transmission. Such individuals, frequently termed 'super-spreaders' (4), are ideal candidates for targeted infection control interventions, including use of limited isolation facilities or novel mucomodulatory approaches to render them rapidly non-infectious. However, we do not know what host and bacillary characteristics drive 'super-spreader' status and disease transmission. We therefore lack the information necessary to rapidly identify infectious individuals or to devise novel solutions. Cough aerosol sampling is a novel technology developed in the United States that has previously been used to quantify the numerical and size distribution of individual cough aerosol particles from TB patients. Unlike methods that rely on animals, such as guinea pigs, it is relatively simple and practical to perform and allows for isolation of bacilli from individual aerosol particles. Our data indicate that the quantity of bacilli in cough droplets is the stronges predictor of recent infection in household contacts of active TB cases (submitted for publication; NEJM), and can therefore serve as a proxy marker of infectiousness. This technology has not been applied in South Africa or to patients with XDR-TB. Our overarching objective is to understand the pathobiology of transmission in patients with DR-TB. We seek to: (i) apply cough aerosol sampling to smear-positive patients with MDR-TB and XDR-TB and confirm that, similar to DS-TB, a minority have culturable TB bacilli in their cough aerosol; (ii) study differences in bacterial (transcriptional and 'lazy persister' phenotype, strain type and point mutations) and host factors (duration and severity of TB disease, cough strength and volume, sputum viscosity, HIV status, chest imaging characteristics) in those patients that are cough aerosol culture-positive versus culture-negative. We will use this information to better understand the genesis of transmission and, if appropriate, generate a preliminary clinical prediction rule which can be used to rapidly identify disease 'super-spreaders'. This tool would require further refinement and validation in future prospective studies. The possible long-term impact of the adoption of such a rule would be a marked reduction in the person-to-person transmission of TB.
Drug-resistant tuberculosis (TB) is major public health disaster and the person-to-person transmission of disease, caused by a significant minority of patients, is the main driver of new cases. This project will use cough aerosol sampling technology (hitherto unvalidated in South Africa and in extensively drug resistant (XDR)-TB) to determine which patients with multi drug-resistant-TB and XDR-TB have live TB bacilli in their cough aerosol, and are hence most likely to spread disease. It will examine how host clinical and bacterial factors differ according to the presence of these bacilli, and will allow for the development of interventions suited to resource-poor settings that can be used to identify the small proportion of patients that spread most of the disease.