Tuberculosis (TB) is the biggest infectious cause of death worldwide, and the biggest cause of death in Sub-Saharan Africa among HIV-positive patients. There is need for a non-sputum-based rapid triage test that identifies individuals with presumptive TB requiring confirmatory diagnostic investigation. Such a test could reduce the burden on health systems, expedite referral and confirmatory testing, and treatment thereby reducing transmission.

A non-sputum triage test is needed as many symptomatic patients including those with HIV, can often not produce high quality sputum (which most current diagnostics rely on). Several blood transcriptional diagnostic signatures produced due to immune responses to M. tuberculosis infection have previously been described, however there is lack of real-world performance data especially in high TB/HIV-endemic African settings where rates of HIV (that could compromise sensitivity) and previous TB (that could compromise specificity) are high. Furthermore, by building on prior research that used untargeted sequencing approaches to identify candidate signatures, the investigators are now at a stage to perform the targeted signature measurement at a large scale and cost-efficient manner as part of prospective diagnostic accuracy analyses in real-world settings.

Using the framework provided by ongoing funded diagnostic clinical trials with similar eligibility criteria, in presumptive TB participants (≥18 years) who visit selected health facilities in Cape Town, South Africa, RADIANT has a unique opportunity to pursue several important research questions. RADIANT aims are to 1) evaluate the sensitivity and specificity of selected concise peripheral host transcriptional signatures for active TB among symptomatic persons in South Africa; 2) design a cost-optimised diagnostic algorithm based on transcriptional signatures, SeroSelectTB (an EDCTP funded pan-African evaluation of a point-of-care serological triage test for active TB) results, and confirmatory bacteriological testing, and 3) characterise bacteriologically-negative patients classified as non-TB to determine if those with elevated host transcriptional signatures have other respiratory pathogens (detected in nasopharyngeal swabs using a commercial multiplex panel) and/or develop active TB within six months (incident active TB).

Timely and correct diagnosis of TB is vital for control of the global TB epidemic. TB screening for passive case-finding relies on symptoms, and its limited by its suboptimal sensitivity, hence there is need for a rapid triage test that identifies symptomatic individuals prior to confirmatory diagnostic investigation. Such a test could reduce the burden on health systems and patients, and expedite referral, confirmatory testing, and treatment. Generally, suitable non-sputum tests are not available, and this is primarily because of a lack of biomarkers. A non-sputum-based triage test is required as many people with TB, including those co-infected with HIV and those with incipient early disease, can often not produce high quality sputum, which most current diagnostics (sputum-based smear microscopy, Mycobacteria Growth Indicator Tube (MGIT) 960 liquid culture, and molecular detection by Xpert MTB/RIF (Xpert) or Xpert MTB/RIF Ultra (Ultra)) rely on.

In patients undergoing pulmonary and/or extrapulmonary TB assessments, a diagnostic test based on blood biomarkers would be optimal, as blood is easily obtainable. Several blood transcriptional diagnostic signatures have previously been described, and these biomarkers hold the most promise in the near future for meeting World Health Organization triage test target product profile criteria. There, however, is lack of real-world prospective performance data especially in high TB/HIV-endemic African settings.

One recent study, which used patients recruited by the group in which this proposal is based, provided the first prospective, systematic head-to-head comparison of the diagnostic accuracy of 27 blood transcriptional signatures for active TB identified through a previous systematic review. This study allowed concise signatures to be identified with Sweeney3, Kaforou25, Roe3 and BATF2 signatures having the highest diagnostic accuracy. This study, however, is limited in its sample size and did not comprehensively characterise patients without TB who had a positive signature score. Therefore, further large-scale, prospective validation studies to compare multiple transcriptional signatures in real-world settings, are important to advance transcriptional signatures through the diagnostics pipeline. Put simply, test developers need to know which, of the signatures that are now available and relatively concise, hold the most promise.

Importantly, these signatures are reflective of the inflammatory state of the host. Furthermore, patients who have elevated signatures but are culture-negative may have sub-clinical incipient TB, in that they will shortly become culture-positive and develop incident active TB. In other words, in such a situation, the signatures could have prospective predictive value and enable clinics to detect patients before they represent a significant infectious risk. As blood transcriptional changes may predate microbiological confirmation, delay in starting treatment pending microbiological culture leads to increased risk of MTB transmission. A study that included longitudinal sampling of HIV-negative adolescents with latent TB revealed that blood transcriptional signatures could detect subclinical incipient disease up to 12 months before a conventional clinical diagnosis was made.

Also, patients who are signature positive but do not have active TB (classically defined culture-positive sputum) are still sick and require characterisation. This could inform future algorithms that investigate such patients. Therefore, the investigators will apply a next-generation sequencing -based respiratory pathogen panel to characterize respiratory pathogens.

As high costs could be a limitation to the implementation of transcriptional signatures as a test for active TB in resource-constrained settings, a strategy to select patients with increased probability of having TB, to get screened by RNA-based triage test before confirmatory testing may result in a more affordable diagnostic algorithm, and improve the cost-effectiveness of RNA-based TB test. The investigators will explore combinations of sensitivity, and specificity of each pre-defined transcriptional signature with other TB tests, including the novel SeroSelectTB blood triage test, to detect which optimal test combinations will improve affordability of the RNA-based assay. A previous study had earlier employed this strategy for the optimal use of the Xpert test.

This proposal therefore aims to evaluate the diagnostic accuracy of selected host transcriptional signatures in the blood of symptomatic participants seeking care at selected health facilities in Cape Town, South Africa, and characterise bacteriologically negative patients classified as non-TB to determine if elevated host transcriptional signatures are associated with other respiratory pathogens (detected in nasopharyngeal swabs using a commercial multiplex panel) or can predict active TB within six months (incident active TB).

The investigators will use, as a scaffold, ongoing funded diagnostic clinical trials with similar eligibility criteria and recruiting participants (≥18 years) with respiratory symptoms who visit selected health facilities in Cape Town, South Africa, to collect Tempus Blood RNA and nasopharyngeal swab specimen at baseline, and longitudinally, in a subset of people.

The study's hypotheses are 1) selected blood transcriptional signatures have a high diagnostic accuracy, defined as minimum of 90% sensitivity and 70% specificity for WHO target product profile for a triage test, when compared to a microbiological reference standard; 2) In bacteriologically negative patients classified as non-TB, baseline host transcriptional signatures can predict active TB within six months (incident TB); (3) respiratory pathogens other than TB in patients bacteriologically negative for TB are associated with elevated transcriptional signatures.

Ultimately, RADIANT aims to show that blood transcriptional signatures can be used as a triage test for TB or as an add-on confirmatory TB test, and characterisation of respiratory tract pathogens has the potential to direct appropriate therapy and infection control precautions.

Inclusion Criteria:

• Adults aged 18 years and above
• Signed written informed consent or witnessed oral consent in case of illiteracy, before undertaking any study-related activities
• Are unwell and are suspected to have tuberculosis

Exclusion Criteria:

• Currently receiving TB treatment
• In the past 3 months, participants have been on TB treatment for 30 or more days

Adults 18 years of age and older who self-report at the selected healthcare facilities in Cape Town, South Africa, seeking medical attention due to symptoms suggestive of tuberculosis within the parent SeroSelectTB study and consent to RADIANT.

Individually identifiable data (demographic, clinical and laboratory) accumulated during the project will be stored on a password-protected electronic database, REDCap. Data will be anonymized, and patient names or addresses will not be stored in the database. Patients are assigned a study number which will be used for study documentation, and annual progress reports.

Transcriptional profiling datasets generated during the current study will be stored in a publicly available repository such as the European Bioinformatics Institute and National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) data repositories, concurrent with the publication of manuscripts.

Data will be shared one year after study completion.

Anyone who is interested in accessing the data can communicate with Dr Anna Okunola via email (annaojo@sun.ac.za) and they will be given access to anonymise data upon reasonable request, after the publication of the main study manuscript. The applicant will also need to sign a Data Transfer Agreement (DTA) with Stellenbosch University.