Can Lipoarabinomannan Antigen Enable Point-of-Care TB Diagnosis? Exploring the Nexus of Clinical Concentration Ranges, Accuracy, and Diagnostic Yield

Background and research questions: This cumulative doctoral dissertation aims to investigate whether urinary Lipoarabinomannan (uLAM) is a suitable biomarker for tuberculosis (TB) diagnosis in general populations. The uLAM-based Alere Determine TB LAM test (Abbott, USA) is currently the only test meeting operational characteristics of the Target Product Profile of the World Health Organization (WHO) and being recommended to assist in the TB diagnosis of patients with HIV co-infection. Its use in general populations is currently not recommended due to the low sensitivity of the test. At the core of this dissertation are three underlying and interconnected research questions, that enable the development and use of uLAM-based tests: (1) What are the clinical concentration ranges of Lipoarabinomannan (LAM) in urine samples in general (non-HIV) patient populations? (2) What is the diagnostic yield of urine LAM tests compared to sputum-based tests in people living with human immunodeficiency virus (HIV)? (3) How can diagnostic yield be incorporated into policy development for novel TB tests? Methods and findings: The three publications address these research questions: (1) A clinical multi-centre diagnostic accuracy study in South Africa and Peru that measured LAM concentrations in urine samples from HIV-negative outpatients using a highly sensitive electrochemiluminescence assay and comparing performance to two point-of-care uLAM tests, the existing Alere Determine TB LAM, and the novel Fujifilm Silvamp TB LAM tests (Fujifilm, Japan). The study found that 66.7% of people with TB without HIV co-infection had uLAM concentrations ≥5 pg/ml when using the highly sensitive electrochemiluminescence assay. Fujifilm Silvamp TB LAM, with a cut-off of 10-20 pg/ml, came close to the Target Product Profile (TPP) and identified 53.2% of positive TB cases, representing a 5-fold increase in sensitivity among HIV-negative patients compared with AlereLAM at 10.8% sensitivity. (2) A systematic review and meta-analysis of individual participant data from existing TB studies employing Bayesian random-effects and mixed-effects meta-analyses to predict the diagnostic yield of urine LAM tests and sputum-based TB tests in people living with HIV. The study, based on 20 datasets with 10202 participants, found that nearly all (98%) participants provided urine, and only 82% provided sputum within 2 days. Due to the lower availability of sputum, the diagnostic yields of sputum Xpert MTB/RIF (Cepheid, USA) (61%) and sputum smear microscopy tests (32%) were lower than their sensitivities. By contrast, the diagnostic yield of urine Alere Determine TB LAM (41%) and Fujifilm Silvamp TB LAM tests (65%) were unaffected by sample provision as urine was readily obtained from almost all people living with HIV. (3) A health policy review and guidance paper that emphasized the importance of diagnostic yield as a measure of the utility of novel TB tests, especially those that use non-sputum specimens. This paper provides clear definitions and examples of diagnostic yield, highlights its relevance for different TB populations and settings, and proposes ways to incorporate and evaluate diagnostic yield in the World Health Organization (WHO) guideline development process. Discussion: The discussion provides an overarching update on 15 LAM research and development areas that form a research strategy and bring the three publications of this dissertation into context of the agenda. After a decade of dedicated research following my LAM strategy, I and others made progress on all 15 research and development areas, paving the way for a next-generation LAM-based TB diagnostic test. The discussion concludes that LAM is a suitable marker for TB diagnosis in general populations and a future test can meet the TPP. However, a critical success factor and possible pitfall is the development of a simple-enough, rapid uLAM point-of-care (POC) test with an extremely low cut-off of around 5 pg/ml, which is required to reach sufficient sensitivity in an unselected population of people with presumed TB. Even if a test does not reach the required sensitivity of 65%, modelling suggests to prioritize uLAM test development, as the benefits of highly available urine-based point-of-care tests can be substantial and tests can reach high population coverage as long as it is low cost. Conclusion: LAM is one of the most promising tuberculosis diagnostic biomarkers, and significant progress showing its potential has been made. These advancements suggest that ultra-sensitive uLAM detection could form the basis for the first non-sputum TB test meeting the World Health Organization's Target Product Profile (TPP), potentially transforming TB diagnosis.