Improving diagnosis of tuberculosis from urine cell-free DNA

Abstract

Tuberculosis (TB) is the leading cause of infectious disease-related mortality worldwide, in part due to limitations in rapid diagnostics. Current TB tests rely on sputum samples, which are difficult to collect from many patients, including people living with HIV, those who are severely ill, and children, and may not detect extrapulmonary TB. Rapid sputum-based tests also have reduced sensitivity in these same patient populations, who more often have paucibacillary TB. There is a critical need for TB diagnostics that target more easily accessible samples, such as urine. Transrenal urine cell-free DNA (cfDNA) is a promising biomarker for TB but is challenging to detect because of the short length (<100 bp) and low concentration of TB-specific fragments. We aimed to increase the sensitivity of TB diagnosis from urine cfDNA by improving recovery of short fragments during the sample preparation step. Here, I present the development of a sequence-specific purification method for cfDNA with <1 copy/mL sensitivity from large volume (10 mL) urine samples that improves detection of short cfDNA relative to silica-based urine cfDNA extraction methods. In a clinical cohort study in South Africa, our assay had 84% sensitivity and 100% specificity for adult pulmonary TB, the highest diagnostic accuracy for a TB urine cfDNA test reported to date. We also characterized urine cfDNA using next-generation sequencing, revealing that TB-derived cfDNA is significantly more degraded than human-derived cfDNA. Finally, I explored several strategies for future assay improvements, including development of a multiplexed assay. The work described in this dissertation offers several important contributions to the growing evidence for urine cfDNA as a diagnostic target for TB, as well as to the broader cfDNA field. Collectively, it motivates continued development of urine cfDNA-based assays for TB diagnosis and will aid informed design of improved cfDNA assays. Our sequence-specific purification approach for urine cfDNA has the potential to expand access to rapid TB diagnostics by enabling diagnosis across sputum-scarce and paucibacillary populations, addressing an urgent need that was identified as one of the highest priority gaps in TB diagnostics.