Researchers Develop Fast, Accurate Tuberculosis Test That Requires No Sputum Sample

Tuberculosis killed 1.25 million people in 2023, reclaiming its position as the world's deadliest infectious disease [1]. That same year, an estimated 4 million cases went undiagnosed — many because the standard method for detecting TB requires something millions of patients simply cannot produce: sputum [2].

On March 9, 2026, the World Health Organization issued its first-ever recommendations endorsing tongue swab specimens and near-point-of-care (NPOC) molecular tests for TB diagnosis [3]. Days later, the New England Journal of Medicine published results from a seven-country trial of a handheld, battery-powered device called the MiniDock MTB that can detect tuberculosis from a tongue swab in roughly 30 minutes — for about $4 per test [4].

The convergence of these two events represents the most significant shift in TB diagnostics in over a decade. But whether this technology can survive the transition from clinical trial to the crowded, underfunded health systems where TB thrives is far from certain.

The Burden: 10.8 Million Cases, Millions Missed

The WHO's 2024 Global Tuberculosis Report documented an estimated 10.8 million new TB cases worldwide in 2023, with 8.2 million formally diagnosed — the highest detection number since WHO began monitoring in 1995 [1]. In 2024, diagnoses rose again to 8.3 million, while deaths fell slightly to 1.23 million [5].

Source: WHO Global TB Report 2024

Data as of Oct 29, 2024CSV

Eight countries account for two-thirds of global cases: India (26%), Indonesia (10%), China (6.8%), the Philippines (6.8%), Pakistan (6.3%), Nigeria (4.6%), Bangladesh (3.5%), and the Democratic Republic of Congo (3.1%) [1]. Southeast Asia alone accounts for nearly 40% of TB deaths [6].

The gap between estimated and diagnosed cases — roughly 2.6 million people in 2023 — is driven in large part by diagnostic limitations. Sputum-based testing, the backbone of TB detection for over a century, fails systematically in three populations: children under 5 (who have low bacterial loads and often cannot cough up phlegm), people living with HIV (whose TB symptoms overlap with other opportunistic infections), and patients with extrapulmonary TB, where the disease affects organs other than the lungs [7]. Between 25% and 40% of symptomatic patients cannot produce a sputum sample at all; among asymptomatic individuals, that figure reaches 90% [8].

Children face particularly acute challenges. Bacteriological test sensitivity drops sharply in young children, and culture yield for pediatric extrapulmonary TB is typically below 50% [7]. An estimated 662,000 TB cases in 2023 occurred in people co-infected with HIV, where diagnostic overlap makes clinical assessment unreliable [1].

How the New Tests Work

The MiniDock MTB, manufactured by Pluslife Biotech (China), is a handheld nucleic acid amplification test (NAAT) — a category of molecular diagnostics that detect the genetic material of Mycobacterium tuberculosis directly. The device consists of a card reader ($180) and a thermolyze unit ($180), with each test card costing approximately $4 [4]. It runs on battery power, requires no laboratory infrastructure, and delivers results in about 30 minutes [9].

The clinical trial, led by Dr. Adithya Cattamanchi of the University of California, Irvine, and funded by the U.S. National Institutes of Health, enrolled 1,380 patients across sites in India, Nigeria, the Philippines, South Africa, Uganda, Vietnam, and Zambia [4]. Using sputum samples, the MiniDock MTB achieved 85.7% sensitivity and 97.6% specificity. Using tongue swabs, sensitivity was 79.6% with specificity of 99.5% [4].

These numbers meet WHO's target product profile for NPOC molecular tests: at least 85% sensitivity and 98% specificity for sputum, and at least 75% sensitivity and 98% specificity for non-sputum specimens [3].

A second technology, ActCRISPR-TB developed at Tulane University by researchers Zhen Huang and Tony Hu, uses a CRISPR-based detection method with multi-guide RNA Cas12a. Published in Nature Communications in 2025, the "one pot" test involves adding a sample to a pre-loaded tube, incubating it, and reading colored bands — a process taking about 45 minutes [10]. On tongue swabs, ActCRISPR-TB reached 74% sensitivity; on respiratory samples, 93%; on pediatric stool samples, 83%; and on adult spinal fluid, 93% [10].

Source: Multiple Sources / WHO

Data as of Apr 30, 2026CSV

Sensitivity and the Bacterial Load Question

The critical comparison point for any new TB diagnostic is the GeneXpert MTB/RIF system, manufactured by Cepheid (a Danaher subsidiary). GeneXpert has been the WHO-recommended rapid molecular test since 2010. On sputum from smear-positive patients (those with high bacterial loads visible under a microscope), GeneXpert achieves 98-99% sensitivity and 99-100% specificity [11]. For smear-negative patients — those with lower bacterial loads — sensitivity drops to a range of 74-96%, depending on the study population [11].

Traditional sputum smear microscopy, while cheap and widely deployed, has sensitivity of only 30-70%, with one large study reporting 48% sensitivity and 84% specificity [11]. Mycobacterial culture remains the gold standard but requires 2-8 weeks to yield results — a timeline that is clinically useless for acute treatment decisions [11].

The MiniDock MTB's 85.7% sputum sensitivity sits slightly below GeneXpert's overall average of approximately 88.6% [11]. The tongue swab figure of 79.6% represents a meaningful trade-off: lower sensitivity in exchange for a sample type that vastly more patients can provide. A separate multicenter study published in the Journal of Infection reported tongue swab concordance of 95.1% with Xpert MTB/RIF, and sensitivity of 88.6% with specificity of 98.3% against a microbiological reference standard [12].

Accuracy degrades predictably at lower bacterial loads. Smear-negative patients — who represent a large share of HIV-co-infected and pediatric cases — are where all molecular tests perform worst. Whether the MiniDock MTB and ActCRISPR-TB maintain adequate sensitivity in these paucibacillary (low-bacteria) populations across diverse real-world settings is the central unanswered question.

The Cost Equation

Economics may prove more decisive than sensitivity in determining whether these tests reach the patients who need them.

Source: MSF / WHO / NEJM

Data as of Apr 30, 2026CSV

A GeneXpert cartridge currently costs $7.97 for low- and middle-income countries (LMICs), reduced from $9.98 in 2023 and $16.87 at its original price [13]. But the real-world cost per test averages $21, ranging from $16 to $58 depending on testing volume, because the total includes instrument maintenance, calibration, and operator time [14]. The four-module GeneXpert instrument itself costs approximately $17,000 [14].

Médecins Sans Frontières (MSF) and over 150 civil society organizations have campaigned for Cepheid to drop all GeneXpert test prices to $5 in LMICs, arguing that production costs are only $3-$5 per cartridge and that the technology was developed with more than $250 million in taxpayer-funded R&D [13].

The MiniDock MTB's economics are markedly different. At $4 per test card with an instrument cost of roughly $360 (card reader plus thermolyze unit), the total capital investment is less than 2.5% of a GeneXpert system. A 2026 analysis in eClinicalMedicine by researchers at Boston University and the University of Amsterdam found that for decentralized molecular testing to be cost-effective, instrument prices need to fall to $400-$800 — a range the MiniDock MTB already meets [15].

"The cost-effectiveness of decentralized testing is overwhelmingly driven by how much the instrument costs and how heavily it gets used," said Dr. Brooke Nichols, one of the study's authors [15]. The same analysis estimated that decentralized molecular testing could diagnose up to 23% more TB cases than current approaches [15].

The Sputum-Free Track Record: Reasons for Caution

Enthusiasm for non-sputum TB diagnostics has been tempered before. The most prominent precedent is the urine lipoarabinomannan (LAM) test, which detects a component of the TB bacterial cell wall in urine.

The original Alere Determine TB LAM test achieved pooled sensitivity of only 42% with 92% specificity [16]. In HIV-negative populations, sensitivity collapsed to 10-18% [16]. The test failed to meet WHO's target product profile, which requires at least 65% sensitivity and 98% specificity regardless of HIV or CD4 status [16].

Newer versions have improved but remain inconsistent. The S4-20 urine LAM assay reported 62% sensitivity and 99% specificity overall, but showed paradoxically higher sensitivity among HIV-negative participants (41%) than HIV-positive patients with CD4 counts above 200 (20%) [17]. The FujiLAM and EclLAM assays have demonstrated significant lot-to-lot and plate-to-plate variation in diagnostic accuracy, with the EclLAM limit of detection ranging from 3 to 63 pg/mL across different plates [18].

These failures highlight several specific risks that regulators should evaluate for tongue swab-based tests:

• Sample degradation: Long-term storage and transport conditions may affect analyte stability, as documented with urine LAM [17].
• Lot-to-lot manufacturing variation: Inconsistent test performance between production batches undermined real-world LAM reliability [18].
• Population-specific performance gaps: Sensitivity that varies dramatically by HIV status or bacterial load creates clinical uncertainty about when to trust a negative result [16].
• Implementation barriers beyond accuracy: Budget limitations, supply chain fragility, and healthcare worker training gaps were the most commonly cited constraints in LAM adoption, independent of test performance [16].

Funding, Patents, and Access

The MiniDock MTB seven-country trial was funded by the U.S. National Institutes of Health [4]. The device is manufactured by Pluslife Biotech, a Chinese diagnostics company. Details on patent ownership, licensing terms, and any equity stakes held by the trial investigators have not been fully disclosed in the published literature.

The ActCRISPR-TB technology was developed at Tulane University School of Medicine [10]. University-developed diagnostics typically involve institutional patent holdings, with licensing terms negotiated separately for commercial and humanitarian use.

Access frameworks such as TB REACH (funded by the Government of Canada and administered by the Stop TB Partnership) and the Global Drug Facility have historically served as mechanisms for subsidizing and distributing TB diagnostics in high-burden countries [19]. Whether Pluslife Biotech or Tulane would participate in these frameworks — accepting volume-based pricing in exchange for market access — remains to be negotiated.

The precedent set by GeneXpert is instructive: despite substantial public R&D funding, Cepheid maintained pricing above production cost for years, prompting sustained civil society pressure [13]. Any new entrant with a $4 price point at launch would face pressure to maintain or reduce that price as production scales.

Regulatory Pathway and Timeline

WHO's March 2026 recommendations on NPOC tests and tongue swabs provide a policy framework but do not constitute prequalification of any specific product [3]. In December 2024, WHO granted its first-ever prequalification to the Xpert MTB/RIF Ultra — after more than a decade of the original Xpert's use [20]. In July 2025, WHO prequalified the Cy-Tb skin test (manufactured by the Serum Institute of India) for TB infection detection [21].

WHO launched a pilot parallel prequalification and policy assessment process for new TB in-vitro diagnostics in 2025, with a call for submissions that closed on August 30, 2025 [22]. Seven additional TB tests are currently under review [22].

For the MiniDock MTB, the regulatory path likely involves:

1. Submission to WHO prequalification, which requires manufacturing site inspection, product dossier review, and independent performance evaluation.
2. Phase III evidence in endemic settings beyond the published seven-country trial, particularly with larger sample sizes in pediatric and HIV-co-infected populations.
3. National regulatory approvals in target countries, which may proceed independently of or in parallel with WHO prequalification.

Given that the Xpert Ultra took years to achieve prequalification despite extensive global use, a realistic timeline for MiniDock MTB WHO prequalification is likely 2-4 years — meaning widespread programmatic deployment before 2028-2030 is unlikely unless expedited pathways are activated.

Market Disruption and Infrastructure Inertia

National TB programs in high-burden countries have built diagnostic networks around sputum collection, GeneXpert platforms, and centralized laboratory systems. India alone has deployed over 4,600 GeneXpert machines through its National TB Elimination Programme [23]. Indonesia, Nigeria, and the Philippines have made similar, smaller-scale investments.

Only 3.9 million of the estimated 10.8 million TB cases in 2023 received molecular testing [15]. The remainder were diagnosed through smear microscopy or clinical assessment alone. A transition to tongue swab-based NPOC testing could reach patients in primary care clinics and community health settings where GeneXpert machines are absent — but it would also cannibalize existing GeneXpert utilization, reducing the per-test throughput that makes those instruments cost-effective.

Cepheid/Danaher, the dominant player in TB molecular diagnostics, would face direct competitive pressure. GeneXpert's installed base is its competitive moat: thousands of instruments in the field, trained operators, established supply chains. But if NPOC devices at $360 can match 80-86% of GeneXpert's diagnostic accuracy at half the per-test cost and a fraction of the instrument price, the economic logic of continued GeneXpert expansion weakens considerably.

Dr. Tereza Kasaeva, director of WHO's department for HIV, TB, Hepatitis, and STIs, framed the stakes bluntly: "These new WHO recommendations mark a major step forward in making TB testing faster and more accessible" [3]. The question is whether the health systems that need these tests most can absorb them before another 4 million annual cases slip through the diagnostic net.

What Remains Unknown

Several gaps in the evidence deserve transparency:

• Long-term field performance: The MiniDock MTB trial ran across seven countries but enrolled 1,380 patients — a modest sample for a disease with 10.8 million annual cases across heterogeneous populations [4].
• Pediatric-specific data: Neither the MiniDock MTB nor ActCRISPR-TB trial has published large-scale results specifically in children under 5, the population where sputum-free testing is most needed.
• Drug resistance detection: The MiniDock MTB detects M. tuberculosis presence but its capacity for rifampicin resistance detection — a core feature of GeneXpert — has not been prominently reported.
• Durability in field conditions: Battery-powered devices in tropical, humid, and resource-limited environments face reliability challenges that controlled trials may not capture.
• Intellectual property terms: Licensing arrangements that would govern pricing in LMICs under TB REACH or Stop TB Partnership frameworks have not been publicly disclosed.

The history of TB diagnostics is littered with promising technologies that faltered on the path from trial to implementation. Tongue swab-based NPOC tests have stronger clinical data behind them than urine LAM tests did at a comparable stage, and their cost profile is genuinely competitive. But the distance between a published NEJM paper and a working diagnostic in a rural clinic in Bihar or Lagos remains vast — and that distance is measured in funding, political will, and years.

TB has killed approximately 1 billion people over the past two centuries. The tools to stop it have never been closer to adequate. Whether they arrive in time depends less on the science than on the systems built to deliver it.