Revisiting diagnostics: high priority tuberculosis diagnostic tests that fill an unmet need: what we need and what we do not need

With 10.8 million incident cases in 2023, the number of individuals developing tuberculosis (TB) is currently at a historical peak [1]. The WHO estimates a case-detection gap of approximately 24% (2.6 million undetected cases), predominantly because of a lack of diagnostics adapted to settings where the majority of TB cases occur and the appearance of new cases because of Mycobacterium tuberculosis transmission and reactivation [1]. There is an urgent need for substantial improvements in the prevention and early diagnostics of TB. To improve prevention, novel tests are needed that predict the risk of TB better than, e.g. tuberculosis skin tests (TSTs) and blood-based interferon-γ release assays (IGRAs). Specific, sensitive, and affordable sputum-free point-of-care (POC) diagnostics would be a major advancement to close the case-detection gap, and such accurate and inexpensive near-POC technologies could potentially also predict M. tuberculosis drug resistance to assure that patients receive effective therapies. New methods are also needed to improve treatment monitoring and to individualize the management of patients. Parallel to the need for improvements, some of the existing diagnostics must be applied more responsibly to save resources and to avoid misinterpretations.

TB is probably the leading cause of death attributed to a single microorganism worldwide. According to the WHO, there were 8.2 million individuals identified with incident TB in 2023 [1]. However, the WHO estimates that 10.8 million individuals developed TB during that year, suggesting a case-detection gap of 2.6 million persons, equivalent to 24% of estimated cases escaping detection [1].

The number of persons with contagious TB is likely higher as emerging evidence indicates that more than 60% of individuals with viable M. tuberculosis from sputum culture have no cough at all and that more than 20% of those persons with TB but without any cough have detectable acid-fast bacilli on sputum microscopy as indicators of higher contagiousness [2]. Without a substantial scale-up of diagnostic capacities, the TB case-detection gap will not be closing any time soon (Fig. 1).

Fig. 1.

Diagnostics for tuberculosis. Tests or techniques that are needed and that can be discarded. DR-TB, drug-resistant TB; DST, drug-susceptibility testing; IGRAs, interferon-γ release assays; POC point of care; PET/CT, positron emission tomography/computed tomography; TB, tuberculosis; TST, tuberculosis skin tests.

Prevention of TB by a vaccine that is more effective than the century-old M. bovis Bacille Calmette Guérin, could have a substantial impact on the global incidence of TB [3]. Nevertheless, as accurate biomarkers of protective immunity remain elusive, TB vaccine development is challenged. There is broad consensus that even when infection with M. tuberculosis occurs, the majority of those who are infected do not develop the disease [4]. Thus, developing TB in those exposed to M. tuberculosis is actually the exception and not the rule (only ~1–2% of exposed persons develop active TB in the short term). Unfortunately, there is currently no test to differentiate those who will eventually fall ill after being exposed and those who will remain healthy. Our limited understanding of mechanisms contributing to protective immunity has been frustrating. Even the best TB vaccine candidate in clinical development, M72:AS01_E, does not offer substantially more protection from TB than food supplements in high-burden settings [5] where more than 100 individuals need to be vaccinated to prevent one case of TB [6].

The most commonly used diagnostic test for M. tuberculosis infection, the tuberculin skin test, developed by Pirquet [7] in 1907 as an allergy probe for the diagnosis of active TB, is even older than the Bacille Calmette Guérin vaccine. Similar to the 21st century blood test versiondthe IGRAsdthe skin tests are unable to distinguish latent infection from active disease on an individual level. Because of the sub-optimal diagnostic accuracy (poor sensitivity and specificity), the use of skin tests or IGRAs has close to zero value for the diagnosis of active TB in adults, a common misunderstanding in clinical practice [8]. By contrast, children have normally not been infected with M. tuberculosis, but if they are infected, they have a high risk of progression to severe TB disease. Therefore, guidelines do recommend the use of TST/IGRAs as a tool to support medical decision making in children [9].

Although the use of IGRA or TST is widely recommended in clinical management guidelines in high-income and some middle-income countries to treat presumed M. tuberculosis infection in the absence of active disease, the positive predictive values of both tests are ~2% in adult contacts, meaning that 98% of those with a positive test result did not progress to active disease [10]. Tests that can determine the presence of viable M. tuberculosis in humans would be highly welcome to target treatment for TB infection more efficiently. Several transcriptomic signatures have been found that promise to predict future incident TB in high-burden settings, but they have not proven to be game changers in terms of their performance metrics [11].

Currently, available methods for diagnosing active TB primarily rely on obtaining a sputum sample from affected individuals. This remains challenging even among symptomatic adults and is particularly difficult in children and the elderly. Moreover, by definition, it is not feasible in cases of extrapulmonary TB without concurrent pulmonary involvement. However, new sampling methods are emerging. New diagnostics are being investigated and might find a place in future clinical practice such identification of M. tuberculosis DNA from tongue or oral swabs, blood or stool or of specific bacillary compounds from exhaled breath or urine [12]. Although a urine Lipoarabinomannan (LAM)-based POC is available, its sensitivity and specificity in HIV-infected and uninfected persons remain sub-optimal. For all other populations, especially for children and adults with paucibacillary pulmonary diseases or patients with extrapulmonary TB, a POC test is still missing. In addition, in low-incidence countries of TB, where diseases by non-tuberculous mycobacteria species are emerging and TB is declining tests that rapidly distinguish non-tuberculous mycobacteria species from M. tuberculosis would be a welcome addition to the methodological spectrum.

A POC test, or near-POC test, for the identification of complex drug-susceptibility profiles of M. tuberculosis, is also urgently needed, especially for high-burden countries with higher rates of drug-resistant TB. Currently, many laboratories in these countries face challenges in meeting the WHO’s target product profiles for molecular tests that predict drug susceptibility. These guidelines recommend a minimum diagnostic sensitivity of more than 95% for detecting rifampicin resistance, more than 90% for isoniazid and fluoroquinolone resistance, and at least 80% for resistance to bedaquiline, linezolid, clofazimine, delamanid, pretomanid, amikacin, and pyrazinamide, when using phenotypic drug-susceptibility testing as the reference standard [13]. Recently, the Xpert ‘XDR’-TB cartridge has been marketed for near-POC testing for isoniazid, fluoroquinolone, aminoglycoside, and ethambutol resistance. If released before 2021, this cartridge would be named correctly; however, considering the new XDR-TB definition, this cartridge is now a ‘preXDR-TB’ cartridge, as resistance against group A drugs other than fluoroquinolones (bedaquiline and linezolid) is not ascertained. Because rifampicin-monoresistance is uncommon and aminoglycosides and ethambutol are not anymore considered first or second-line choices for the treatment of drug-resistant TB, the added value of the cartridge boils down to the detection of fluoroquinolone resistance and isoniazid monoresistance. Given that the regimen of choice to treat multidrug-resistant/ridampicin-resistant-tuberculosis (MDR/RR-TB) is bedaquiline–pretomanid–linezolid–moxifloxacin, the practical relevance of this cartridge is very limited.

Treatment of TB could potentially be improved by biomarkers identifying patients who would benefit most from host-directed therapies, such as N acetylcysteine or everolimus, aiming to decrease post-TB lung disease and to improve long-term outcomes. In addition, TB treatment outcomes could be improved by accurate POC tests to monitor treatment [13]. Ideally, TB treatment should be guided by information that accurately predicts (a) adverse events of therapies before and during the selection of treatment regimens (e.g. suprabasin), (b) M. tuberculosis culture conversion (e.g. TB27), and (c) individual timepoint to guide treatment termination to secure that relapse-free cure has been achieved (e.g. TB22 and R/S ratio).

Currently, available tools fail to accurately monitor treatment effectiveness. IGRAs are unsuitable for this purpose as positive M. tuberculosis-specific immune responses frequently persist despite successful treatment. Most patients with pulmonary TB who are cured have ongoing inflammation at the end of therapy as documented by positron emission tomography/computed tomography (PET/CT) and is thus not a useful imaging marker to guide the duration of therapy [14]. MRI could have similar properties but has not been explored in detail for TB treatment monitoring. Similarly, sequential measures of M. tuberculosis-specific DNA from sputum by Xpert or other nucleic acid amplification tests are not recommended because of DNA persistence unlinked to cure. Advances in technologies, such as the mycobacterial load assay, or the R/S ratio assay, distinguish viable from dead bacteria. Biomarkers in combination with clinical features (weight and nutritional indices), imaging characteristics using standardized scoring of radiological disease extent and cavity scores (e.g. TIMIKA) [15], mycobacterial disease burden, and host biomarker profiles hold some promise for treatment monitoring, especially if used as a validated bio-clinical score in the future.

Despite the limitations of current TB diagnostics, access to newer tools, such as automated low-complexity nucleic acid amplification tests (e.g. GeneXpert), remains limited in many high-burden settings. Therefore, identifying effective mechanisms for the rapid adoption and accessibility of newly developed TB diagnostics in areas where they are most needed is crucial to ensuring that these tools have a meaningful impact on TB control.

In summary, despite major diagnostic advancements in recent years, substantial obstacles remain to predict which M. tuberculosis-exposed individuals will progress to active TB, to diagnose active TB when bacteria are not readily available from sputum, and to guide the treatment of people affected by different clinical manifestations of TB.

New TB diagnostics should be adapted to resource-limited settings to have a global impact on TB control. Without improvements in TB diagnostics and widespread implementation of diagnostic advances, the case-detection gap will continue to exist (and potentially even grow), hindering efforts to eliminate TB.