A stool based qPCR for the diagnosis of TB in children and people living with HIV in Uganda, Eswatini and Mozambique (Stool4TB): a protocol for a multicenter diagnostic evaluation

Tuberculosis (TB) remains a major cause of morbidity and mortality worldwide, resulting in 10.6 million incident cases and 1.6 million deaths in 2021 [1]. The African region is one of the most TB-affected areas in the world, representing approximately 23% of the global TB disease burden, 26% of deaths among HIV negative people and 73% of the deaths among people living with HIV (PLHIV) [2]. Both children and PLHIV have an increased risk of TB progression and mortality, particularly in the absence of rapid diagnosis [1, 3]. Recent modeling studies suggest that around 96% of children dying from TB do not receive adequate treatment and from those under 5 years of age 80% are not even diagnosed [4]. Early and appropriate TB diagnosis in these two populations is a key pillar of the WHO End TB Strategy, where specific emphasis is placed on the discovery, development, and rapid uptake of new diagnostic tools and effective strategies for their implementation and scale-up [5].

Despite some advances in TB diagnostics in recent years, such as the emergence of Xpert Ultra and LF-LAM, diagnosing and confirming TB remains a challenge particularly among children and immunocompromised PLHIV. In these groups, TB often presents without traditional TB clinical symptoms, [6, 7] making passive case detection based on symptom screening less effective at detecting TB. The paucibacillary nature of disease [8] in children and PLHIV, and challenges in obtaining respiratory samples due to difficulty spontaneously expectorating sputum also create obstacles towards bacteriological confirmation. Thus, new diagnostic strategies, based on the use of easy-to-collect non-invasive specimens, could improve clinical decision making and TB outcomes TB high burden settings.

Since children and PLHIV are less likely to produce quality sputum, increased interest has been placed on exploring new specimens for TB diagnosis and sputum free diagnostics was listed as one of the WHO’s high priority target products for TB diagnostics [9]. Recent studies have identified stool as an effective child-friendly specimen, since airway secretions are normally cleared into the digestive system. These studies went on to inform the most recent WHO guidelines, which now recommend stool as an initial sample for microbiological diagnosis in children [10].

Despite its increasing availability in high TB burden settings, the 4^th generation Xpert® Ultra in sputum (Cepheid Inc., Sunnyvale, CA, USA) confirms TB in 90% of adult cases with HIV-infection[11] and 62–89% of culture-positive pediatric cases [12]. While mycobacterial culture is the accepted reference standard for assessing novel TB diagnostic tests, it only confirms TB in 10–50% of children clinically diagnosed with pulmonary TB [13‐16] and about half of PLHIV starting on anti-tuberculosis treatment (ATT).

To address this diagnostic gap, researchers at Baylor College of Medicine (Houston, Texas, USA) developed a novel, stool bead-based, real-time quantitative PCR (qPCR) diagnostic test for TB. A pilot study evaluating this qPCR have shown its ability to increase TB case confirmation among adults in Eswatini and children in Tanzania with clinically diagnosed, bacteriologically negative TB (by sputum Xpert) by 12% and 19%, respectively. This study also points to the test’s ability to predict TB treatment failure [17]. In Eswatini, the detection of M.tb DNA in the stool after 2 months of ATT was associated with treatment failure, death, or DR-TB (RR 3.5, 95% CI: 1.32- 9.64; p = 0.015) [17]. Thus, this qPCR platform provides an important opportunity to explore the utility of a qPCR for treatment monitoring via quantification of M.tb burden in stool. In addition, this test could be combined with novel approaches to assess M. tb viability using molecular mycobacterial load assays [18].

Stool Xpert MTB/RIF was reported to have a limit of detection (LOD) greater than 267 CFU/mL [19]. In contrast, when using a bead-based DNA isolation kit designed for soil (MP Fast DNA, MPBio), the LOD was lowered to 96 Mtb Colony Forming Units (CFU) per 50 mg of stool (95% CI 84–105 CFU), which approximately represents 5 CFU/mL [20, 21]. There is still little evidence about the performance of Xpert Ultra on stool, but a recent Cochrane review reports a sensitivity of around 60% against culture (varying from 39 to 100% including data from unpublished cohorts) [12]. For Xpert Ultra the LOD has been lowered to 16 CFU/mL. This shows some room for improvement, for which we expect the qPCR, combined with the DNA isolation kit developed for stool specimens, to present with a higher sensitivity than Xpert® Ultra. Therefore, we propose a diagnostic evaluation to assess the performance of this stool-based qPCR for TB bacteriological confirmation among PLHIV and children with the hypothesis that it may present an increased sensitivity for TB diagnosis compared to currently recommended microbiological tests.

This study is designed to evaluate a new diagnostic test that aims to improve molecular confirmation of TB as well as provide a new monitoring tool for treatment response in children and PLHIV. To date, studies using stool as a target sample have been done with already existing molecular tests like Xpert and Xpert Ultra using stool, but few of those studies propose novel molecular platforms for validation or exclusively include the vulnerable populations of intended use (children or PLHIV with presumptive TB). Therefore, the relevance of this study relies not only on the use of stool as a sample for bacteriological confirmation, but also on the molecular non-Xpert based approach to it, as there is need for improvement molecular stool diagnostics. Further, the design is aligned with the latest WHO guidelines on diagnosis and management of TB in children and adolescents that, for the first time, include stool as a valid sample for TB testing. In fact, apart from the classical approach of using culture in a respiratory sample as a reference standard, we will assess the performance of this new diagnostic test against a robust composite reference standard that includes culture and Xpert Ultra in respiratory samples, Xpert Ultra in stool and LF-LAM in urine and against clinical diagnosis. In this way, we will both maximize the bacteriological confirmation and estimate the potential increase in bacteriological confirmation in those cases that start treatment in the absence of bacteriologic confirmation. It will also allow to see the contribution of each single TB test to bacteriological confirmation, including the expected additionality of que novel stool-based qPCR (all other test negative but qPCR positive).

Another relevant strength in the study design is the two week and two-month longitudinal follow-up to verify baseline disease classification in participants who do not initiate treatment. This will play a relevant role in both increasing the number of confirmed cases (after resampling if persistence of symptoms), but also in the reliability of the specificity estimations (only those who have tested negative at baseline, who did not start treatment and who do not present symptoms after 2 months will be classified as “unlikely TB”).

The study is designed to include those presumptive TB cases already detected by the routine procedures of the national TB programs or healthcare system workers and does not include those identified in community screening programs. Therefore, recruitment will rely on the performance of the already existing platforms for TB detection. While this is good practice and prevents the study from negatively impacting future clinical practice, the COVID-19 pandemic has impacted TB notifications in many study settings [1]. This may impact recruitment and the capacity to attain the expected sample size.

Though stool has already been recommended by the WHO as a sample for bacteriological confirmation in children, the implementation of stool-based diagnostics has not yet been widespread. Therefore, this study is highly relevant as it uses stool as the preferred sample. Also, the new qPCR platform has the potential to lower the limit of detection of Mtb in stool compared to currently available tests. If our project successfully demonstrates an increase in bacteriological confirmation, this platform has the potential to be adapted to a field-friendly point-of-care test that could be easily implemented in basic health care centers in Sub-Saharan Africa or resource-limited countries, in line with the priorities of the End TB strategy [38]. As a multi-national study with different epidemiological and environmental backgrounds, one can expect to face variability, but this will make results more robust and able to be applied in different environments.